TW201344679A - Multi-resolution switched audio encoding/decoding scheme - Google Patents

Abstract

An audio encoder for encoding an audio signal comprises a first coding branch, the first coding branch comprising a first converter for converting a signal from a time domain into a frequency domain. Furthermore, the audio encoder comprises a second coding branch comprising a second time/frequency converter. Additionally, a signal analyzer for analyzing the audio signal is provided. The signal analyzer, on the hand, determines whether an audio portion is effective in the encoder output signal as a first encoded signal from the first encoding branch or as a second encoded signal from a second encoding branch. On the other hand, the signal analyzer determines a time/frequency resolution to be applied by the converters when generating the encoded signals. An output interface includes, in addition to the first encoded signal and the second encoded signal, a resolution information identifying the resolution used by the first time/frequency converter and used by the second time/frequency converter.

Description

Multi-resolution switching audio encoding/decoding scheme (1) Field of invention

The present invention relates to audio coding, and in particular to low bit rate audio coding schemes.

In the prior art, a frequency domain coding scheme such as MP3 or AAC is known. The frequency domain encoders are based on a time domain/frequency domain conversion, a subsequent quantization phase, wherein in the subsequent quantization phase, information from a sensing module is used to control the quantization error, and an encoding phase, wherein The quantized spectral coefficients and corresponding side information are entropy encoded using a coding table.

On the other hand there are encoders that are very suitable for speech processing, such as AMR-WB+ as described in 3GPP TS 26.290. Such speech coding schemes perform one of the time domain signals, linear prediction (LP) filtering. This LP filtering is derived from linear prediction analysis of one of the input time domain signals. The resulting LP filter coefficients are then quantized/encoded and transmitted as side information. This process is called Linear Predictive Coding (LPC). At the output of the filter, the predicted residual signal or prediction error signal, also referred to as the excitation signal, is encoded using a comprehensive analysis stage of the ACELP encoder, or alternatively using a conversion encoder that has an overlap. Fourier transform. A closed loop or an open loop algorithm is used to determine the excitation coding (also known as TCX coding) using ACELP coding or transcoding.

A frequency domain audio coding scheme, such as an efficient AAC (HE-AAC) coding scheme that combines an AAC coding scheme with a band replica (SBR) technique, or a stereo or a multi-voice associated with one of the so-called "MPEG Surround" The channel coding tool is combined.

On the other hand, a speech encoder such as AMR-WB+ also has a high frequency extension stage and a stereo function.

An advantage of the frequency domain coding scheme is that they display a high quality for low bit rate music signals. However, the speech quality of the low bit rate is problematic.

The speech coding scheme shows high quality even for low bit rate speech signals, but shows poor quality for other signals at low bit rates.

Summary of invention

It is an object of the present invention to provide a method of solving the above mentioned problems and an apparatus for carrying out the method. The object of the present invention is to provide an improved encoding/decoding concept by the method described in the scope of the independent patent application of the present invention.

This object is achieved by an audio encoder according to item 1 of the patent application scope, an audio coding method according to claim 9 of the patent application scope, a decoder according to claim 10, and claim 19 according to the patent application scope. A decoding method is implemented according to one of the coded ones of claim 20 or a computer program according to claim 21 of the scope of the patent application.

The present invention is based on the discovery that a coding/encoding scheme of a hybrid or dual mode switch has the advantage that an optimal coding algorithm can always be selected for a certain signal characteristic. In other words, the present invention does not expect a signal encoding algorithm that perfectly matches all signal characteristics. This solution will always be a compromise, by the conventional audio encoder and voice editing The huge difference between the encoders can be appreciated. Instead, the present invention combines different encoding algorithms, such as a speech encoding algorithm and an audio encoding algorithm, in a switching scheme to select the best matching encoding algorithm for each portion of the audio signal. Moreover, the two encoding branches include a time/frequency converter but provide a further domain converter such as an LPC processor in an encoding branch, which is also a feature of the present invention. The domain converter confirms that the second branch is adapted to a certain signal characteristic than the first encoding branch. However, the conversion of the signal output of the domain processor to a spectral representation is also a feature of the present invention.

Two converters, ie the first converter in the first encoding branch and the second converter in the second encoding branch, are configured to implement a multi-resolution conversion encoding, wherein The audio signal and the resolution of the corresponding converter are specifically set according to the audio signal actually encoded in the corresponding encoding branch to obtain a good compromise between quality and bit rate or to take into account A fixed quality, lowest bit rate or a fixed bit rate, the highest quality.

In accordance with the present invention, the time/frequency resolution of the two converters can preferably be set independently of one another such that each time/frequency converter can optimally match the time/frequency resolution requirement of the corresponding signal. The bit efficiency, that is, the relationship between the useful bit and the side information bit, is higher for the longer block size/window length. Therefore, it is preferred that the two converters are biased toward a longer window length because substantially the same amount of side information relates to one of the audio signals compared to the application of a shorter block size/window length/conversion length. Part of the longer time. Preferably, the time/frequency resolution in the encoding branches is also affected by other encoding/decoding tools located in these branches. Preferably, the second encoding branch comprising the domain converter, such as an LPC processor, includes another hybrid scheme, such as an ACELP branch and a TCX scheme, wherein the second converter is included in the TCX scheme. Better The resolution of the time/frequency converter located in the TCX branch is also affected by the encoding decision such that a portion of the signal in the second encoding branch is in the TCX having the second converter Processing in the branch or in the ACELP branch without a time/frequency converter.

Basically, the domain converter and the second encoding branch, and in particular the first processing branch in the second encoding branch and the second processing branch in the second encoding branch, are not necessarily speech related The components, such as an LPC analyzer of the domain converter, a TCX encoder of the second processing branch, and an ACELP encoder of the first processing branch. Other applications are also useful when other signal characteristics of an audio signal different from voice and music are evaluated. Any domain converter and coding branch implementation can be used, and a best matching algorithm can be found using a comprehensive analysis scheme such that at the decoder side all coding selections are performed for each portion of the audio signal and the best results are selected, wherein The best result can be found by implementing an objective function on the encoded results. Next, identifying (for a decoder) side information of the basic encoding algorithm for a portion of the encoded audio signal is coupled to the encoded audio signal through an encoder output interface such that the decoder does not have to Any decision on the encoder side or on any signal characteristics is chosen to select its coding branch based only on the side information of the transmission. In addition, the decoder will not only select the correct decoding branch, but also select which time/frequency resolution will be applied to a corresponding first decoding branch and a corresponding one based on the side information encoded in the encoded signal. In the second decoding branch.

Accordingly, the present invention provides an encoding/decoding scheme that combines the advantages of all of the different encoding algorithms to avoid the disadvantages of these encoding algorithms, when the signal portion would have to be unsuitable for a certain encoding algorithm These shortcomings occur when encoding one of the algorithms. Moreover, the present invention avoids the need to say that different time/frequency resolution requirements due to different portions of the audio signal in different encoding branches have not been said Any shortcomings that will arise in the future. Instead, due to the variable time/frequency resolution of the time/frequency converters in the two branches, at least a reduction or even a complete avoidance will be used in this case: ie at the same time/frequency resolution will be used for the two codes Any artifacts that occur in the case of a branch or where any coded branch will only be a fixed time/frequency resolution.

The second switch is again determined between the two processing branches, but in a different domain than the "external" first branch domain. Again, an "internal" branch is initiated primarily by a source model or SNR calculation, and other "internal" branches can be initiated through a mask, and/or a perceptual model, or through at least a frequency/spectrum. Domain coding level. Illustratively, an "internal" branch has a frequency domain encoder/spectrum converter and the other branch has one of the encoders encoded on other domains, such as the LPC domain, wherein the encoder is, for example, not performing One of the input signals, CELP or ACELP quantizer/scaler, is processed in the case of a spectral conversion.

A further preferred embodiment is an audio encoder comprising a first information slot directional to an encoding branch such as a spectral domain encoding branch, and an encoding branch oriented such as an LPC domain encoding branch a second information slot source or SNR and a switch for switching between the first encoding branch and the second encoding branch, wherein the second encoding branch includes a specific domain different from the time domain a converter (such as an LPC analysis stage that generates one of the excitation signals), and wherein the second coding branch additionally includes a specific domain such as an LPC domain processing branch and a specific spectrum such as an LPC spectral domain processing branch An additional switch for the domain and for switching between the particular domain encoding branch and the particular spectral domain encoding branch.

A further embodiment of the present invention is an audio decoder comprising a first domain (such as a spectral domain decoding branch) and a second domain (such as for decoding one in the second domain) a signal (such as an excitation signal), one of the LPC domain decoding branches) and a third domain (such as for decoding) An LPC spectral decoder branch of a signal (such as an excitation signal) in a third domain (such as an LPC spectral domain), wherein the third domain is performed by performing a frequency conversion from the second domain Obtaining, wherein a first switch is provided for the second domain signal and the third domain signal, and wherein the first domain decoder is provided between the first domain decoder and the decoder for the second domain or the third domain Switch one of the second switches.

3, 4‧‧ items

3a, 3b, 3c, 3d‧‧‧ frames

10a‧‧‧LPC analysis level

60‧‧‧Long-term predictive component

62‧‧‧Short-term forecasting component

64‧‧‧ Codebook

66, 68‧‧‧ blocks

69, 86‧‧‧ subtractor

70‧‧‧All-pole filter

72‧‧ ‧ glottal model

77‧‧‧ Gain level

78‧‧‧Transfer path

79‧‧‧Reward path

80‧‧‧Additional level

81, 85‧‧‧ predictive filter

84‧‧‧ line

87, 88, 89, 411, 536‧ ‧ blocks

99‧‧‧ audio input signal

100‧‧‧Common preprocessing stage

101‧‧‧Combined multi-channel, surround/combined stereo

101a‧‧‧ downmixer

101b‧‧‧Joint Multichannel Parameter Analyzer

102‧‧‧width extension

102a‧‧‧High Band Analyzer

102b‧‧‧Low-pass filter block

195‧‧‧Intermediate signal

200, 521‧‧ ‧ switch

300, 525‧‧‧Signal analyzer, decision stage, controller

300a‧‧‧Signal Analyzer

400‧‧‧First Coded Branch

410‧‧‧ first converter

410a, 523a‧‧‧ windowizer

410b, 523b‧‧‧ converter

421‧‧‧Quantizer/Scaler Block, Quantizer/Encoder Level, Spectrum Audio Encoder

424‧‧‧Quantizer/Scaler Block

431‧‧‧First decoding branch, spectrum audio decoder, decoding/dequantization stage

440‧‧‧First Decoder Branch, Controllable Converter, First Domain Converter, Time Domain Converter, Frequency/Time Converter

440a, 534a‧‧‧inverter stage

440b, 534b‧‧‧Synthetic window level

440c, 534c‧‧‧Overlap/Additional

450‧‧‧First decoding branch

500‧‧‧Second code branch

510‧‧‧LPC processor, domain converter

522‧‧‧Quantizer/Scaler Block, First Processing Branch

523‧‧‧Second converter

524‧‧‧ Further coding tools

526‧‧‧ACELP Core, ACELP Block, ACELP Time Domain Encoder

527‧‧‧TCX block, MDCT-TCX processing unit

531‧‧‧First inverse processing branch, component, inverse quantizer/encoder, decoding/dequantization stage

532‧‧‧Combiners, components, switches

533‧‧‧Second reverse processing branch, inverse quantizer/encoder, decoding/dequantization stage, term

534‧‧‧Second reverse processing branch, inverse spectrum converter, frequency/time converter, components

537‧‧‧Transport, TCX ^-1 block

540‧‧‧First converter, domain converter, LPC synthesis stage, decoding branch

550‧‧‧Second decoding branch

600‧‧‧ combiner, switch

601‧‧‧ mode detection block, mode decision

609‧‧‧Decoding audio signals

699‧‧‧Decoding audio intermediate signal

701‧‧‧Bandwidth expansion block, bandwidth extension level

701a‧‧‧Take machine

701b‧‧‧ adjuster

701c‧‧‧ combiner

702‧‧‧Joint stereo/surround processing level, joint multi-channel level, item

702a‧‧‧Decoder end, upmixer

702b‧‧‧Decoder end, parameter decoder

799‧‧‧Decoding audio signals

800‧‧‧Output interface, bit stream multiplexer

801‧‧‧Encoder output signal

900‧‧‧Input interface, bit stream demultiplexer

1200‧‧‧Psychoacoustic Module

1201‧‧‧Time domain noise trimming tools and components

1202‧‧‧M/S coding tools and components

1203‧‧‧M/S decoding tool

1204‧‧‧TNS decoder tool

1205‧‧‧ bass back filter

Simple illustration

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The preferred embodiments of the present invention are described with respect to the accompanying drawings, wherein: FIG. 1a is a block diagram of one of the first aspects of the present invention, and FIG. 1b is the first in accordance with the present invention. a block diagram of one of the decoding schemes of the layer; FIG. 1c is a block diagram of one of the coding schemes according to one of the further aspects of the present invention; and FIG. 2a is a block diagram of one of the coding schemes according to one of the second aspects of the present invention Figure 2b is a schematic diagram of one of the decoding schemes of the second layer according to the present invention; Figure 2c is a block diagram of one of the coding schemes according to one of the further aspects of the present invention; a block diagram of one of the coding schemes of one of the further layers; FIG. 3b illustrates a block diagram of one of the decoding schemes of the further layer in accordance with the present invention; and FIG. 3c illustrates the encoding apparatus/method having a cascade switch a schematic representation; Figure 3d illustrates a schematic diagram of one of the devices or methods for decoding (in which a cascade combiner is used); Figure 3e illustrates one of the time domain signals illustrating and illustrating the short interleaved fade-out regions included in the two encoded signals. a corresponding representation of one of the encoded signals; Figure 4a illustrates a block diagram of a switch positioned prior to the encoding branch; and Figure 4b illustrates a coding scheme having one of the switches positioned after the encoding branch a block diagram; Figure 5a illustrates one of the time domain speech segments as one of the quasi-periodic or similarly pulsed signal segments; Figure 5b illustrates one of the segments of Figure 5a; and Figure 5c illustrates the silent speech One time domain speech segment as an example for a similar noise segment; Figure 5d illustrates one of the time domain beams of Figure 5c; Figure 6 illustrates a block diagram of a comprehensive analysis CELP encoder; The 7d diagram illustrates an audible/silent excitation signal as an example for a similar pulse signal; Figure 7e illustrates the provision of short-term prediction information and one of the prediction error (excitation) signals at the encoder-end LPC stage; Produce A further embodiment of one of the LPC devices of a weighted signal; Figure 7g illustrates a reversal weighting operation and a subsequent excitation analysis required by implementing the converter 537 as in Figure 2b Converting the weighted signal into one of the excitation signals; Figure 8 illustrates a block diagram of a joint multi-channel algorithm in accordance with one embodiment of the present invention; Figure 9 illustrates a preferred embodiment of a bandwidth extension algorithm; Figure 10a illustrates a detailed description of one of the switches when performing an open loop decision; and Figure 10b illustrates the operation of the file in a closed loop decision mode When one of the switches is illustrated.

11A is a block diagram of an audio encoder according to another aspect of the present invention; FIG. 11B is a block diagram showing another embodiment of an inventive audio decoder; and FIG. 12A is a diagram showing another embodiment of an inventive encoder; Embodiment 12B illustrates another embodiment of an inventive decoder; FIG. 13A illustrates the relationship between resolution and window/conversion length; and FIG. 13B illustrates a group conversion window for the first encoding branch An overview and transition from one of the first to the second encoding branches; Figure 13C illustrates a plurality of different window sequences, including a window sequence for the first encoding branch and one for the second branch a sequence of transitions; Figure 14A illustrates the framing of a preferred embodiment of the second encoding branch; Figure 14B illustrates a short window applied to the second encoding branch; and Figure 14C illustrates the application to the second a window of equal size in the encoding branch; Figure 14D illustrates the long window applied by the second encoding branch; Figure 14E illustrates the ACELP in a hyperframe division An exemplary sequence of frames and TCX frames; Figure 14F illustrates different conversion lengths corresponding to different time/frequency resolutions for the second encoding branch; and Figure 14G illustrates multiple uses of Figure 14F Define one of the windows to construct.

Figure 11A illustrates an embodiment of an audio encoder for encoding an audio signal. The encoder includes a first encoding branch 400 for encoding an audio signal using a first encoding algorithm to obtain a first encoded signal.

The audio encoder further includes a second encoding branch 500 for encoding an audio signal using a second encoding algorithm to obtain a second encoded signal. The first encoding algorithm is different from the second encoding algorithm. Additionally, a first switch is provided for switching between the first encoding branch and the second encoding branch such that the first encoded signal or the second encoded signal is for a portion of the audio signal In an encoder output signal 801.

The audio encoder illustrated in FIG. 11A additionally includes a signal analyzer 300/525 configured to analyze a portion of the audio signal to determine the portion of the audio signal at the encoding The output signal 801 is represented as the first encoded signal or the second encoded signal.

The signal analyzer 300/525 is further configured to variably determine one of the first converters 410 in the first encoding branch 400 or one of the second converters 523 in the second encoding branch 500. One of each time/frequency resolution. This time/frequency resolution is applied when the first encoded signal or the second encoded signal representing the portion of the audio signal is generated.

The audio encoder additionally includes an output interface 800 for generating the encoder output signal 801. The encoder output signal 801 includes an encoded representation of the portion of the audio signal and indicates the audio signal. The representation is the first encoded signal or the second encoded signal and indicates one of the time/frequency resolutions used to decode the first encoded signal and the second encoded signal.

The second encoding branch is preferably different from the first encoding branch in that the second encoding branch additionally includes means for converting the audio signal from its processed domain in the first encoding branch to a One of the different domain domain converters. Preferably, the domain converter is an LPC processor 510, but the domain converter can be implemented in any other manner as long as the domain converter is different from the first converter 410 and the second converter 523.

The first converter 410 is a time/frequency converter, preferably including a windower 410a and a converter 410b. The windower 410a applies an analysis window to the input audio signal, and the converter 410b performs the conversion of the windowed signal to a spectral representation.

Similarly, the second converter 523 preferably includes a windower 523a and a subsequently connected converter 523b. The windower 523a receives the signal output of the domain converter 510 and outputs its windowed representation. The result of an analysis window applied by the windower 523a is input to the converter 523b to form a spectral representation. The converter can be implemented as a FFT or preferably an MDCT processor in a software or hardware or a hybrid hardware/software implementation. Alternatively, the converter may be a filter bank implementation, such as a QMF filter bank, which may be based on a real-valued or complex modulation of a prototype filter. For a particular filter bank implementation, a window is applied. However, for other filter bank implementations, a windowing required for a conversion algorithm based on an FFT or MDCT is not necessary. When using a filter bank implementation, then the filter The wave set is a variable resolution filter bank and the resolution controls the frequency resolution of the filter bank and additionally the temporal resolution or only the frequency resolution without the temporal resolution. However, when the converter is implemented as an FFT or MDCT or any other corresponding converter, then the frequency resolution is connected to the time resolution due to one of the frequency resolutions obtained for a larger block length. The increase automatically corresponds to a lower time resolution in time, and vice versa.

Additionally, the first encoding branch can include a quantization/encoder stage 421, and the second encoding branch can also include one or more further encoding tools 524.

Importantly, the signal analyzer is configured to generate a resolution control signal for the first converter 510 and the second converter 523. Therefore, an independent resolution control is implemented in both encoding branches to have one encoding scheme that provides a low bit rate on the one hand and one of the highest quality considering one of the low bit rates on the other hand. In order to achieve this low bit rate target, longer window lengths or longer conversion lengths are preferred, but where these long lengths will result in an artifact (due to low temporal resolution), the application results in a lower frequency resolution. Shorter window length and shorter conversion length. Preferably, the signal analyzer applies a statistical analysis or any other analysis suitable for the corresponding algorithms in the encoding branches. In the implementation mode where the first coding branch is a frequency domain coding branch (such as an AAC-based encoder) and the second coding branch includes an LPC processor 510 as a domain converter, the signal The analyzer performs a speech/music distinction to control the switch 200 such that the voice portion of the audio signal is fed into the second encoding branch. The music portion of the audio signal is fed to the first first encoding branch 400 by correspondingly controlling the switch 200 indicated by the switch control lines. Alternatively, the switch can also be positioned at the output interface 800 as will be discussed later with respect to FIG. 1C or FIG. 4B. The front.

Additionally, the signal analyzer can receive an audio signal input to the switch 200 or an audio signal output by the switch 200. Additionally, the signal analyzer performs an analysis to not only feed the audio signal to the corresponding encoding branch, but also to determine the appropriate time/frequency resolution of the respective converter in the corresponding encoding branch. The first converter 410 and the second converter 523 are indicated, for example, by a resolution control line connected to the signal analyzer and the converter.

Figure 11B contains a preferred embodiment of an audio decoder that matches one of the audio encoders of Figure 11A.

The audio decoder in Figure 11B is configured to decode an encoded audio signal, such as the encoder output signal 801 output by the output interface 800 in Figure 11A. The encoded signal includes a first encoded audio signal encoded according to a first encoding algorithm, and a second encoded signal encoded according to a second algorithm (the second encoding algorithm is different from the first encoding algorithm) And indicating whether the first encoding algorithm or the second encoding algorithm is used to decode information of the first encoded signal and the second encoded signal and for one of the first encoded audio signal and the second encoded audio signal Time/frequency resolution information.

The audio decoder includes first decoding branches 431, 440 for decoding the first encoded signal based on the first encoding algorithm. Additionally, the audio decoder includes a second decoding branch for decoding the second encoded signal using the second encoding algorithm.

The first decoding branch includes a first controllable converter 440 for converting from a spectral domain to one of the time domains. The controllable converter is configured to control the first decoded signal using the time/frequency resolution information from the first encoded signal.

The second decoding branch is included for converting from a spectral representation One time represents one of the second controllable converters, and the second controllable converter 534 is configured to be controlled using the time/frequency resolution information 991 for the second encoded signal.

The decoder additionally includes a controller 990 for controlling the first converter 540 and the second converter 534 based on the time/frequency resolution information.

Additionally, the decoder includes a domain converter for generating a composite signal using the second decoded signal to cancel the domain conversion applied by the domain converter 510 in the encoder of FIG. 11A.

Preferably, the domain converter 540 is an LPC synthesis processor controlled using LPC filter information included in the encoded signal, wherein the LPC filter information has been generated by the LPC processor 510 in FIG. 11A. And as side information has been input to the encoder output signal. The audio decoder finally includes a combiner 600 for combining the first decoded signal output by the first domain converter 440 with the composite signal to obtain a decoded audio signal 609.

In the preferred implementation, the first decoding branch additionally includes a dequantizer/decoder stage 431 for reversing or at least partially reversing one of the operations performed by the corresponding encoder stage. However, it is clear that quantification is not reversed because it is a loss operation. However, a dequantizer will reverse some of the non-uniformity in one of the quantization such as one-to-one or over-extension quantization.

In the second decoding branch, the corresponding stage 533 applies for some encoding operations imposed by the cancellation stage 524. Preferably, stage 524 includes a uniform quantization. Therefore, the corresponding stage 533 will not have a particular dequantization level for canceling one of the uniform quantizations.

The first converter 440 and the second converter 534 can include a corresponding inverse converter stage 440a, 534a, a composite window level Overlap/addition stages 440c, 534c of 440b, 534b and subsequent connections. Such overlap/addition stages are required when the converters and, more particularly, the converter stages 440a, 534a implement aliasing lead-in conversions such as a modified discrete cosine transform. This overlap/add operation will then perform a time domain aliasing cancellation (TDAC). However, when the converter applies a non-aliased lead-in conversion such as an inverse FFT, then an overlap/add phase 440c is not required. In this implementation, it may be applied to avoid one of the block artifacts being staggered out.

Similarly, the combiner 600 can be a switch combiner or a staggered fade out combiner or when aliasing is used to avoid block artifacts, a combiner is implemented by the combiner, similar to one in it. An overlap/addition level within the branch.

Figure 1a illustrates an embodiment of the invention having two cascade switches. A mono signal, a stereo signal or a multi-channel signal is input to the switch 200. The switch 200 is controlled by the decision stage 300. The decision stage receives a signal as an input for input to block 200. Optionally, the decision stage 300 can also receive side information included in the at least one signal, the stereo signal or the multi-channel signal or at least associated with the signal, in the presence of, for example, when the order is initially generated In the case of an acoustic signal, the stereo signal, or the information generated when the multi-channel signal is generated.

The decision stage 300 activates the switch 200 to feed a signal to the LPC domain encoding portion 500 as illustrated in the frequency encoding portion 400 illustrated in one of the branches of Figure 1a or in the lower branch of Figure 1a. A key component of the frequency domain encoding branch is the spectral conversion block 410, which is operative to convert a common pre-processing stage output signal (as discussed below) into a spectral domain. The spectral conversion block may comprise an MDCT algorithm, a QMF, an FFT algorithm, a Wavelet analysis or a filter bank, such as a key sampling filter bank having a certain number of filter bank channels, Where the sub-bands in the filter bank The signal can be a real value signal or a complex value signal. The output of the spectral conversion block 410 can be encoded using a spectral audio encoder 421, as is known from the AAC encoding scheme. The spectral audio encoder 421 can include processing blocks.

In general, the processing in branch 400 is processed in one of a perception-based model or an information slot model. Therefore, this branch simulates the human auditory system receiving sound. In contrast, the processing in branch 500 is used to generate a signal in the excitation, residual or LPC domain. In general, the processing in branch 500 is a process in a speech model or an information generation model. For speech signals, this model is a model of a human speech/sound generation system that produces sound. However, if a sound from a different source requires a different sound generation model to be encoded, the processing in branch 500 may be different.

In the lower encoding branch 500, a key component is an LPC device 510 that outputs LPC information for controlling one of the characteristics of an LPC filter. This LPC information is transmitted to a decoder. The output signal of the LPC stage 510 is an LPC domain signal composed of an excitation signal and/or a weighted signal.

The LPC device generally outputs an LPC domain signal, which may be any signal in the LPC domain, such as the excitation signal in Figure 7e or a weighted signal in Figure 7f or by The LPC filter coefficients apply to any other signal that has been generated by an audio signal. In addition, an LPC device can also determine these coefficients and can also quantize/code these coefficients.

The decision in the decision stage may be signal adaptive such that the decision stage performs a music/speech distinction and is input to the upper branch 400 as a music signal and the speech signal is input to one of the lower branches 500. The way to control the switch 200. In an embodiment, the decision level will decide it The policy information is fed to an output bit stream such that a decoder can use the decision information to perform the correct decoding operation.

This decoder is illustrated in Figure 1b. The signal output by the spectral audio encoder 421 is input to a spectral audio decoder 431 after transmission. The output of the spectral audio decoder 431 is input to a time domain converter 440. Similarly, the output of the LPC domain encoding branch 500 of Figure 1a is received at the decoder side and processed by elements 531, 533, 534 and 532 to obtain an LPC excitation signal. The LPC excitation signal is input to an LPC synthesis stage 540 that receives the LPC information generated by the corresponding LPC synthesis stage 510 as a further input. The output of the time domain converter 440 and/or the output of the LPC synthesis stage 540 is input to a switch 600. The switch is controlled by a switch control signal, such as generated by the decision stage 300 or externally provided, such as by the original mono, stereo, or multi-channel signal creator. The output of the switch 600 is a complete mono, stereo or multi-channel signal.

The input signal to the switch 200 and the decision stage 300 can be a mono signal, a stereo signal or a multi-channel signal or generally an audio signal. Depending on the decision taken by the switch 200 input signal or from any external source, such as one of the initial audio signals that form the basis of the signal input to stage 200, the switch is at the frequency encoding branch 400. Switching with the LPC encoding branch 500. The frequency encoding branch 400 includes a spectral conversion stage 410 and a subsequently coupled quantization/encoding stage 421. The quantization/encoding stage may include any functionality as known from modern frequency domain encoders, such as AAC encoders. Moreover, the quantization operation in the quantization/encoding stage 421 can be controlled by generating a sensing module, such as a perceptual mask threshold, wherein the information is input to the stage 421.

In the LPC encoding branch, the switch output signal is transmitted through The LPC analysis stage 510, which generates one of the LPC side information and the LPC domain signal, is processed. The excitation encoder creatively includes an additional switch for switching the LPC domain signal between a quantization/encoding operation 522 in the LPC domain or processing quantization/encoding stage 524 in one of the values in the LPC spectral domain. Further processing. For this purpose, a spectral converter 523 is provided at the input of the quantization/encoding stage 524. The switch 521 is controlled in an open loop manner or in a closed loop manner, depending, for example, on the particular settings described in the AMR-WB+ technical description.

For the closed loop control mode, the encoder additionally includes an inverse quantizer/encoder 531 for the LPC domain signal, an inverse quantizer/encoder 533 for the LPC spectral domain signal, and an item 533 for An inverse spectrum converter 534 is output. The encoded and decoded signals in the processing branches of the second encoding branch are all input to the switch control means 525. In the switch control device 525, the two output signals are compared to each other and/or compared to an objective function or an objective function can be calculated based on a comparison of the distortions on the two signals such that the use of the lower distortion is used The signal determines which position the switch should use. In addition, in the case where the two branches provide a non-constant bit rate, the branch providing the lower bit rate may be selected, even when the signal noise of the branch is more than the signal noise of the other branch. When it is low. Additionally, the objective function can use the signal to noise ratio of each signal and one bit rate per signal and/or additional criteria (as an input) to find the best decision for a particular target. If, for example, the goal is to make the bit rate low enough, then the objective function will greatly depend on the bit rate of the two signals output by elements 531, 534. However, when the primary goal is to have the best quality for a certain bit rate, then the switch control 525 may, for example, discard each signal above the allowed bit rate, and when the two signals are in the allowed bit. When the ratio is below the rate, the switch control will select a better signal to noise ratio (ie, with less quantization/coding distortion) signal.

The decoding scheme (as described above) in accordance with the present invention is illustrated in Figure 1b. There is a particular decoding/dequantization stage 431, 531 or 533 for each of the three possible output signal types. When stage 431 outputs a time spectrum, the time frequency spectrum is converted to the time domain using the frequency/time converter 440, stage 531 outputs an LPC domain signal and item 533 outputs an LPC spectrum. To ensure that the input signals to switch 532 are all in the LPC domain, the LPC spectrum/LPC converter 534 is provided. The output data of the switch 532 is converted back to the time domain using an LPC synthesis stage 540, which is controlled by the LPC information generated and transmitted by the encoder. Then, after block 540, both of the branches have time domain information switched according to a switch control signal to finally obtain an audio signal, such as a single sound, which is obtained by inputting the signal into the coding scheme of FIG. 1a. Signal, a stereo signal or a multi-channel signal.

Figure 1c illustrates a further embodiment of a different configuration of the switch 521 having a similar principle to that of Figure 4b.

Figure 2a illustrates a preferred coding scheme in accordance with one of the second aspects of the present invention. A pre-processing scheme common to one of the inputs of the switch 200 can include a surround/join stereo block 101 that produces a joint stereo parameter and a mono output signal as an output, the mono output The signal is generated by downmixing the input signal having two or more channels. In general, the signal at the output of block 101 may also be one of two or more channels, but due to the downmix function of block 101, the number of channels output at block 101 will be greater than the input to the block. The number of channels of 101 is small.

A common pre-processing scheme may include (different from or in addition to the block 101) a bandwidth extension stage 102. In the embodiment of Figure 2a, the output of block 101 is input to the bandwidth extension block 102, In the encoder of Fig. 2a, the bandwidth extension block 102 outputs a signal of a limited frequency band such as a low frequency signal or a low pass signal at its output. Preferably, this signal is also downsampled (e.g., by a factor of two). Furthermore, for signals of the high frequency band input to the block 102, bandwidth extension parameters such as known as the HE-AAC outline from MPEG-4, known as spectral envelope parameters, inverse filtering parameters, noise layer parameters, etc. are generated and Transfer to a meta-stream multiplexer 800.

Preferably, the decision stage 300 receives the signal input to the block 101 or input to the block 102 to make a decision between, for example, a music mode or a speech mode. The branch 400 is encoded on the music mode selection, and the branch 500 is encoded in the voice mode selection. Preferably, the decision stage additionally controls the joint stereo block 101 and/or the bandwidth extension block 102 to adapt the functionality of the blocks to the particular signal. Thus, when the decision stage determines that a certain portion of the input signal is the first mode, such as the music mode, then particular features of block 101 and/or block 102 may be controlled by the decision stage 300. Additionally, when the decision stage 300 determines that the signal is in a speech mode or substantially in a second LPC domain mode, then the particular features of blocks 101 and 102 can be controlled based on the decision level output.

Preferably, the spectral conversion of the encoding branch 400 is accomplished using an MDCT operation, i.e., a more specific warping time operation, wherein the intensity or generally the torsional strength can be controlled between zero and a high distortion strength. In a zero twist strength, the MDCT operation in block 411 is one of the direct MDCT operations known in the art. The time warp strength along with the time warping side information can be transmitted/inputted into the bit stream multiplexer 800 as side information.

In the LPC encoding branch, the LPC domain encoder may include an ACELP core 526 that calculates a pitch gain, a pitch lag, and/or one of the codebook information such as a codebook index and gain. The TCX mode known from 3GPP TS 26.290 causes a perceptually weighted signal in the transition domain One of the treatments. A split multi-rate lattice quantization (algebraic VQ) with noise factor quantization is used to quantize the weighted signal of a Fourier transform. A conversion is calculated in a 1024, 512 or 256 sampling window. The excitation signal is recovered by inverse filtering the quantized weighted signal through a inverse weighting filter.

In the first coding branch, a spectral converter preferably includes a specifically adjusted MDCT operation having certain window functions, a quantization/entropy coding stage consisting of a single vector quantization stage, followed by Preferably, however, it is a scalar quantizer/entropy coder that is combined with one of the quantizer/encoder in the frequency domain coding branch, i.e., item 421 in Fig. 2a.

In the second encoding branch, the LPC block 510 is present, followed by a switch 521, which in turn is followed by an ACELP block 526 or a TCX block 527. ACELP is described in 3GPP TS 26.190 and TCX is described in 3GPP TS 26.290. In general, the ACELP block 526 receives an LPC excitation signal as calculated by a program as described in Figure 7e. The TCX block 527 receives a weighted signal as produced with reference to Figure 7f.

At TCX, the conversion is applied to a weighted signal that is computed by filtering the input signal through an LPC-based weighting filter. The weighting filter used in the preferred embodiment of the invention is given by (1- A ( z / γ )) / (1 - μz ^-1 ). Therefore, the weighted signal is an LPC domain signal and its conversion is an LPC spectral domain. The signal processed by ACELP block 526 is the excitation signal and is different from the signal processed by block 527, but both signals are in the LPC domain.

At the decoder side illustrated in Figure 2b, after the inverse spectral conversion of block 537, the inverse of the weighting filter is applied, i.e., (1- μz ^-1 ) / (1 - A ( z / γ )). The signal is then filtered through (1-A(z)) to enter the LPC excitation domain. Therefore, the conversion to the LPC domain block 534 and the TCX-1 block 537 includes reverse conversion and subsequent filtering. Converting from the weighting domain to the excitation domain.

Although item 510 in the 1a, 1c, 2a, 2c diagram illustrates a single block, block 510 can output different signals as long as the signals are in the LPC domain. The actual mode of block 510, such as the excitation signal pattern or the weighted signal pattern, may depend on the actual switching state. Additionally, the block 510 can have two parallel processing devices, one of which is implemented similar to Figure 7e and the other device is implemented as Figure 7f. Thus, the LPC field at the output of 510 can represent the LPC excitation signal or the LPC weighted signal or any other LPC domain signal.

In the second coding branch (ACELP/TCX) of Fig. 2a or 2c, the signal is pre-emphasized by a filter 1-0.68 z ^-1 prior to encoding. In the ACELP/TCX decoder of Figure 2b, the composite signal is boosted by the filter 1/(1-0.68 z ^-1 ). The pre-emphasis may be part of the LPC block 510 where the signal is pre-emphasized prior to LPC analysis and quantization. Similarly, de-enhancement may be part of the LPC synthesis block LPC-1 540.

Figure 2c illustrates a further embodiment of one of the implementations of Figure 2a, but with a different configuration than the switch 521 of the principle of Figure 4b.

In a preferred embodiment, the first switch 200 (see FIG. 1a or FIG. 2a) is controlled by an open loop decision (as in FIG. 4a) and the second switch is controlled by a closed loop decision ( As shown in Figure 4b).

For example, Figure 2c has the second switch placed after the ACELP and TCX branches as shown in Figure 4b. Next, in the first processing branch, the first LPC field indicates the LPC excitation, and in the second processing branch, the second LPC field indicates the LPC weighting signal. That is, the first LPC domain signal is obtained by filtering through (1-A(z)) to be converted to the LPC residual domain, and the second LPC domain signal is filtered by the filter (1- A ( z) / γ )) / (1 - μz ^-1 ) is obtained by switching to the LPC weighting domain.

Figure 2b illustrates one of the decoding schemes corresponding to the encoding scheme of Figure 2a. The bit stream generated by the bit stream multiplexer 800 of Fig. 2a is input to the one bit stream demultiplexer 900. Depending on, for example, a message obtained via a bitstream of a mode detection block 601, a decoder-side switch 600 is controlled to forward a signal from the upper leg or a signal from the lower leg to the bandwidth extension. Block 701. The bandwidth extension block 701 receives the side information from the bit stream demultiplexer 900 and reconstructs the high frequency band based on the low frequency band output by the switch 600.

The full band signal generated by block 701 is input to the joint stereo/surround processing stage 702 that reconstructs two stereo channels or several multi-channels. In general, block 702 will output more channels than input to this block. Depending on the application, the input to block 702 may even include two channels, such as in a stereo mode, or even multiple channels, as long as the output of this block has more channels than the input to the block. .

The switch 200 has been shown to switch between the two branches such that only one path receives a signal for processing and the other does not receive a signal for processing. In an alternative embodiment, however, the switch can also be configured, for example, after the audio encoder 421 and the excitation encoders 522, 523, 524, which means that the two branches 400, 500 process the same signal in parallel. . In order not to double the bit rate, however, only signals output by one of these encoding branches 400 or 500 are selectively written into the output bit stream. The decision stage will then operate to minimize the signal written to the bit stream by a certain cost function, wherein the cost function can be the generated bit rate or the resulting perceived distortion or a combined ratio / Distortion cost function. Thus, in this mode or in the modes illustrated in the various figures, the decision stage can also operate in a closed loop mode to confirm that only the code is ultimately available. The branch is written to the bit stream that has the lowest bit rate for a given perceptual distortion or the lowest perceptual distortion for a given bit rate. In the closed loop mode, the feed input can be taken from the outputs of the three quantizer/scaler blocks 421, 522, and 424 in Figure 1a.

In an implementation with two switches (ie, the first switch 200 and the second switch 521), it is preferable that the time resolution for the first switch is lower than the time resolution for the second switch. . In other words, the blocks of the input signal to the first switch (switched by a switching operation) are larger than the blocks switched by the second switch operating in the LPC domain. Illustratively, the frequency domain/LPC domain switch 200 can switch blocks of length 1024 samples, and the second switch 521 can switch blocks each having 256 samples.

Although some of the figures 1a through 10b are illustrated as a block diagram of a device, these figures are also an illustration of one of the methods in which a plurality of block functions correspond to a plurality of method steps.

Figure 3a illustrates an audio encoder for generating an encoded audio signal as an output of the first encoding branch 400 and a second encoding branch 500. Moreover, the encoded audio signal preferably includes side information such as pre-processing parameters from the common pre-processing stage or switch control information as discussed with respect to the previous figures.

Preferably, the first encoding branch is operable to encode an audio intermediate signal 195 in accordance with a first encoding algorithm, wherein the first encoding algorithm has an information slot model. The first encoding branch 400 generates a first encoded output signal that is a spectral information representation encoded by one of the audio intermediate signals 195.

In addition, the second encoding branch 500 is adapted to encode the audio intermediate signal 195 according to a second encoding algorithm, the second encoding algorithm having an information source model and the information indicating the intermediate audio signal The source model produces (in a second encoder output signal) the encoded parameters.

The audio encoder further includes a common pre-processing stage for pre-processing an audio signal 99 to obtain the intermediate audio signal 195. In particular, the common pre-processing stage is operative to process the audio input signal 99 such that the audio intermediate signal 195 (i.e., the output of the common pre-processing algorithm) is a compressed version of the audio input signal.

A preferred method for generating an audio code of an encoded audio signal comprises the steps of: encoding a 400-intermediate signal 195 according to a first coding algorithm, the first coding algorithm having an information slot model and generating ( In a first output signal, the spectral information of the encoded audio signal is represented; a step of: encoding a 500-intermediate signal 195 according to a second encoding algorithm, the second encoding algorithm having an information source model and generating ( In a second output signal, a parameter for encoding the information source model of the intermediate signal 195 and a step of: pre-processing 100 an audio input signal 99 to obtain the audio intermediate signal 195, wherein the common pre-processing In this step, the audio input signal 99 is processed such that the audio intermediate signal 195 is a compressed version of the audio input signal 99, wherein the encoded audio signal includes the first output signal or a portion of the audio signal. Second output signal. The method preferably includes the further step of encoding a portion of the intermediate signal of the audio using the first encoding algorithm or using the second encoding algorithm or encoding the signal using the two algorithms, and The result of a coding algorithm or the result of the second coding algorithm is output in an encoded signal.

In general, the audio coding algorithm used in the first encoding branch 400 reflects and simulates the situation in an audio slot. The slot of an audio message is usually the human ear. The human ear can be modeled as a frequency analyzer. Therefore, the first encoding branch outputs the encoded spectral information. Preferably, the first encoding branch further comprises for additionally applying a perception The mask is one of the perception models. The perceptual mask threshold is used when quantizing the audio spectral values, wherein preferably the quantization is performed such that a quantization noise is introduced by quantizing the spectral audio values hidden below the perceptual mask threshold.

The second coding branch represents an information source model that reflects sound generation. Thus, the information source model can include a speech model that is reflected by an LPC analysis stage by converting a time domain signal into an LPC domain and subsequently processing the LPC residual signal (ie, the excitation signal). However, the alternative sound source model is a sound source model for representing an instrument or any other sound generator, such as a particular sound source that exists in the real world. When, for example, based on an SNR calculation, that is, based on the fact that the source model is most suitable for encoding a certain time portion and/or frequency portion of an audio signal, several sound source models can be obtained, and different sound source models can be executed. A choice between the two. Preferably, however, switching between the encoding branches is performed in the time domain, i.e., using a model to encode a certain time portion and encoding another different time portion of the intermediate signal using another encoding branch.

Some parameters are used to represent the information source model. When considering a modern speech coder such as AMR-WB+, as for the speech model, the parameters are LPC parameters and encoded excitation parameters. The AMR-WB+ includes an ACELP encoder and a TCX encoder. In this case, the encoded excitation parameters may be global gain, noise layer, and varying length coding.

Figure 3b illustrates one of the encoders corresponding to the encoder illustrated in Figure 3a. In general, Figure 3b illustrates a decoder for decoding an encoded audio signal to obtain a decoded audio signal 799. The decoder includes the first decoding branch 450 for decoding a signal encoded in accordance with a first encoding algorithm having a slot model. The audio decoder further includes a model for decoding based on having an information source model A second encoding algorithm encodes one of the encoded information signals, a second decoding branch 550. The audio decoder further includes a combiner for combining the output signals from the first decoding branch 450 and the second decoding branch 550 to obtain a combination. The combined signal illustrated in FIG. 3b is input as a decoded intermediate signal to a post-processing stage that is used to post-process the decoded intermediate intermediate signal 699 (the combined signal output by the combiner 600). The output signal of one of the common pre-processing stages is an extended version of the combined signal. Therefore, the decoded audio signal 799 has an enhanced information content as compared to the decoded audio intermediate signal 699. This information extension is provided by the common post-processing stage with the aid of pre/post processing parameters which may be transmitted from an encoder to a decoder or may be taken from the decoded intermediate intermediate signal itself. Preferably, however, the pre/post processing parameters are transmitted from an encoder to a decoder because the program allows for improved quality of one of the decoded audio signals.

Figure 3c illustrates an audio decoder for decoding an audio input signal 195. According to the preferred embodiment of the present invention, the audio input signal 195 can be identical to the intermediate audio signal 195 of Figure 3a. The audio input signal 195 appears in a first domain, which may be, for example, a time domain but may be any other domain, such as a frequency domain, an LPC domain, an LPC spectral domain, or any other domain. In general, the conversion from one domain to another is performed by a conversion algorithm, such as any of the conventional time/frequency conversion algorithms or frequency/time algorithms.

For example, a selectable transition from the time domain to the LPC domain is the result of filtering the LPC of a time domain signal, which causes an LPC residual signal or an excitation signal. As may be the case, any other filtering operation that produces a filtered signal that has an effect on a large number of signal samples prior to conversion can be used as a conversion algorithm. Therefore, weighting an audio signal using an LPC-based weighting filter is a further conversion, and the conversion is A signal is generated in the LPC domain. In a time/frequency conversion, a modification to a single spectral value has an effect on all time domain values prior to conversion. Similarly, modifications to any time domain samples will have an impact on each frequency domain sample. Similarly, a modification to sampling one of the excitation signals in the case of an LPC domain will have an effect on the large number of samples prior to the LPC filtering due to the length of the LPC filter. Similarly, a modification to a sample prior to an LPC conversion will have an impact on many of the samples obtained for this LPC conversion due to the inherent memory effects of the LPC filter.

The audio encoder of Figure 3c includes a first encoding branch 400 that produces a first encoded signal. The first encoded signal may be in a fourth domain. In the preferred embodiment, the fourth domain is the time spectral domain, ie, the domain obtained when a time domain signal is processed via a time/frequency conversion. .

Therefore, the first encoding branch 400 for encoding an audio signal uses a first encoding algorithm to obtain a first encoded signal, wherein the first encoding algorithm may or may not include a time/frequency conversion algorithm.

The audio encoder further includes a second encoding branch 500 for encoding an audio signal. The second encoding branch 500 obtains a second encoded signal using a second encoding algorithm different from the first encoding algorithm.

The audio encoder further includes a first switch 200 for switching between the first encoding branch 400 and the second encoding branch 500 such that a portion of the audio input signal is The first encoded signal output by block 400 or the second encoded signal at the output of the second encoding branch is included in an encoder output signal. Therefore, when a portion of the audio input signal 195 is used, the first encoded signal in the fourth domain is included in the encoder output signal. The first encoded signal in the second domain or the second encoded signal of the second processed signal in the third domain is not included in the encoder output signal. This ensures that this encoder is efficient at bit rate. In an embodiment, any time portion of the audio signal included in the two different encoded signals is smaller than the frame length of a frame as discussed for Figure 3e. In the case of a switching event, these small portions are useful for staggering out from one encoded signal to another to reduce artifacts that may occur without any interleaving. Thus, in addition to the interleaved area, each time domain block is represented by a signal encoded by only one of the single fields.

As described in FIG. 3c, the second encoding branch 500 includes means for converting the audio signal (ie, signal 195) in the first domain to a converter 510 in a second domain. In addition, the second encoding branch 500 includes a first processing branch 522 for processing an audio signal in the second domain to obtain one of the second domains. A processed signal causes the first processing branch 522 to not perform a domain change.

The second encoding branch 500 further includes a second processing branch 523, 524 that converts the audio signal in the second domain to a third domain and processes The audio signal in the third domain obtains a second processed signal at the output of the second processing branch 523, 524, wherein the third domain is different from the first domain and also different from the second domain.

In addition, the second encoding branch includes a second switch 521 for switching between the first processing branch 522 and the second processing branch 523, 524 such that for input to the first A portion of the audio signal in the second encoding branch, the first processed signal in the second domain or the second processed signal in the third domain is in the second encoded signal.

Figure 3d illustrates a decoder for decoding one of the encoded audio signals produced by the encoder of Figure 3c. In addition to a contiguous fading area, the interleaved fading area is preferably shorter than the length of the frame to obtain a system as far as possible at the critical sampling limit, using a second domain signal, a third domain signal Or a fourth domain coded signal to represent each block of the first domain audio signal. The encoded audio signal includes the first encoded signal, a second encoded signal in a second domain, and a third encoded signal in a third domain, wherein the first encoded signal, the second encoded The signal and the third encoded signal all relate to different time portions of the decoded audio signal and for a decoded audio signal, the second domain, the third domain and the first domain are each other different.

The decoder includes one of the first decoding branches for decoding based on the first encoding algorithm. The first decoding branch is illustrated at 431, 440 in Figure 3d and preferably includes a frequency/time converter. The first encoded signal is preferably in a fourth domain and converted to the first domain for the decoded output signal.

The decoder of Fig. 3d further comprises a second decoding branch comprising several elements. These elements are a first inverse processing branch 531 for inverse processing the second encoded signal to obtain one of the first in the second domain at the output of block 531. Reverse processed signal. The second decoding branch further includes a second reverse processing branch 533, 534 for inverse processing a third encoded signal to obtain the second domain And a second reverse processed signal, wherein the second reverse processing branch includes a converter for converting from the third domain to the second domain.

The second encoding branch further includes a first combiner 532 for using the first reverse processed signal with the The second inverse processed signal is combined to obtain a signal in the second domain, wherein the combined signal is only affected by the first reverse processed signal and at a subsequent time instant at the first time instant Only affected by the signal of the second reverse processing.

The second decoding branch further includes a converter 540 for converting the combined signal to the first domain.

Finally, the decoder illustrated in FIG. 3d includes a second combiner 600 for outputting the decoded first signal from blocks 431, 440 and the converter 540. The phases are combined to obtain an output signal that has been decoded in one of the first domains. Moreover, the decoded output signal in the first domain is only affected by the signal output by the converter 540 at the first time instant and is only output by the blocks 431, 440 at a subsequent time instant. The first decoded signal is affected.

This is illustrated in Figure 3e from the perspective of an encoder. The upper portion of Figure 3e illustrates in a schematic representation a first domain audio signal, such as a time domain audio signal, wherein the time index increases from left to right and item 3 can be considered to represent the signal 195 in Figure 3c. A series of audio samples. Figure 3e illustrates frames 3a, 3b, 3c, 3d that are generated by switching between the first encoded signal and the first processed signal and the second processed signal (as described in item 4 of Figure 3e) . The first encoded signal, the first processed signal, and the second processed signal are all in different domains and in order to ensure that switching between the different domains does not cause an artifact at the decoder end, then The frame signals 3a, 3b of the domain signal have an overlap indicating one of the staggered fade areas, and the staggered fade areas are in frames 3b and 3c. However, there is no such interlaced fade-out area between the frames 3d, 3c, which means that the frame 3d is also represented by a second processed signal (ie, a signal in the third domain), and There is no domain change between boxes 3c and 3d. Therefore, in general, it is preferable not to provide a staggered fading in the absence of a domain change, and A staggered fading area is provided when there is a domain change (i.e., a switching action of one of the two switches), i.e., a portion of the audio signal encoded by two subsequently encoded/processed signals. Preferably, the staggered fade is performed for other domain changes.

In embodiments where the first encoded signal or the second processed signal has been generated by an MDCT process having, for example, 50% overlap, each time domain sample is included in both subsequent frames. However, due to the multiple characteristics of the MDCT, this does not result in a burden because the MDCT is a critical sampling system. In this paper, critical sampling means that the number of spectral values is equal to the number of time domain values. An advantage of the MDCT is that it provides a crossover effect without a specific crossover region so as to provide from one MDCT block to one of the next MDCT block without any burden of violating critical sampling requirements. Crossover.

Preferably, the first coding algorithm in the first coding branch is based on an information slot model, and the second coding algorithm in the second coding branch is based on an information source model or an SNR model. An SNR model is not specifically related to a particular sound generation mechanism but can be selected from one of a plurality of coding modes based, for example, on a closed loop decision. Therefore, an SNR model is any available coding model, but it does not necessarily have to have a physical composition of the sound generator, but it is any parameterized coding model that is different from the information slot model and can be determined through a closed loop And specifically by comparing different SNR results from different models.

As illustrated in Figure 3c, a controller 300, 525 is provided. The controller may include a plurality of functions of the decision stage of Figure 1a and additionally may include the functionality of the switching device 525 of Figure 1a. In general, the controller is for controlling the first switch and the second switch in a signal adjustment manner. The controller is operative to analyze an input to the first switch Or a signal obtained by the first or second encoding branch output signal or from an encoding and decoding of the first and second encoding branches for an objective function. Alternatively or additionally, the controller is operative to analyze input to the second switch or input or pass by the first processing branch or the second processing branch from the first processing branch for an objective function Signals obtained by processing and reverse processing of the road and the second processing branch.

In an embodiment, the first encoding branch or the second encoding branch comprises an aliasing introduction time/frequency conversion algorithm, such as one of MDCT or one MDST different from one of the direct FFT conversions that introduces an aliasing effect. Algorithm. In addition, one or two branches include a quantizer/entropy encoder block. Specifically, only the second processing branch of the second encoding branch includes the time/frequency converter that introduces an aliasing operation, and the first processing branch of the second encoding branch includes a quantizer and / or entropy encoder and does not introduce any aliasing effect. The aliasing introduction time/frequency converter preferably includes a windower for implementing an analysis window and an MDCT conversion algorithm. In particular, the windower is operative to apply the window function to the subsequent frame in an overlapping manner such that one of the windowed signals is sampled in at least two subsequent window frames.

In one embodiment, the first processing branch includes an ACELP encoder and a second processing branch includes an MDCT spectrum converter and a quantizer for quantizing the spectral components to obtain quantized spectral components, wherein each quantization The spectral components are zero or defined by one of the plurality of different possible quantizer indices.

Moreover, preferably, the first switch 200 operates in an open circuit manner and the second switch operates in a closed loop manner.

As previously described, the two encoding branches are operable to decode the audio signal in a group mode, wherein the first switch or the second switch is switched in a group mode such that a switching action is predetermined at least in one of the signals number The sampling occurs after one of the blocks, and the predetermined number forms a frame length for the corresponding switch. Therefore, the block used for the first switch switching may be a block of, for example, 2048 or 1028 samples, and the frame length (the first switch 200 is based on its switching) is variable but preferably fixed in such a manner A fairly long cycle.

In contrast, when the second switch 521 is switched from one mode to another mode, the block length for the second switch 521 is substantially smaller than the block length for the first switch. Preferably, the two block lengths for the switch are selected such that the longer block length is an integer multiple of the shorter block length. In the preferred embodiment, the block length of the first switch is 2048 or 1024 and the block length of the second switch is 1024 or preferably 512 and more preferably 256 and more preferably 128 is sampled such that the second switch can be switched up to 16 times when the first switch is switched only once. However, a preferred maximum block length ratio is 4:1.

In a further embodiment, the controllers 300, 525 are operative to perform a speech music distinction for one of the first switches in a manner that determines one of the speech decisions relative to one of the music decisions. In this embodiment, one of the decisions on speech is taken even when less than 50% of the portion of the first switch is voice and more than 50% of the frame is music.

In addition, when a relatively small portion of the first frame is voice and a portion of the first frame that is specifically 50% of the length of the smaller second frame is voice, the controller is operatively Switched to this voice mode. Thus, even when, for example, only 6% or 12% of a block corresponds to the frame length of the first switch, a preferred voice/preference switching decision has been switched to speech.

This procedure is preferably to fully utilize the bit rate saving capability of the first processing branch having a voiced speech core in an embodiment and Even the rest of the large first frame of non-speech does not relax any quality, since the second processing branch includes a converter and is therefore also useful for audio signals having non-speech signals. Preferably, the second process comprises an overlapping MDCT that is critically sampled and provides an efficient and alias-free operation even in small window sizes due to the time domain aliasing cancellation, such as overlap at the decoder end. And add. Furthermore, a large block length for the first coding branch (preferably an AAC-like MDCT coding branch) is useful because the non-speech signal is typically quite stationary and a long conversion window provides a high frequency The resolution and thus the high quality, and additionally, because a perceptually controlled quantization module provides a one-bit rate efficiency, the perceptually controlled quantization module can also be applied in the second processing branch of the second encoding branch The conversion-based coding mode.

In the case of the 3d diagram decoder illustration, preferably, the transmission signal includes an explicit indicator as the side information 4a as illustrated in Figure 3e. The side information 4a is retrieved by a one-bit stream parser not illustrated in FIG. 3d to forward the corresponding first encoded signal, first processed signal or second processed signal to the correct processor The first decoding branch, the first reverse processing branch or the second reverse processing branch, such as in Figure 3d. Therefore, the decoded signal not only has the encoded/decoded signal but also includes side information related to these signals. However, in other embodiments, there may be one implicit communication that allows a decoder end bitstream parser to distinguish between certain signals. In the case of FIG. 3e, it is outlined that the first processed signal or the second processed signal is the output of the second encoding branch and thus the second encoded signal.

Preferably, the first decoding branch and/or the second reverse processing branch comprises switching from the spectral domain to one of the time domain MDCT exchanges. For this purpose, an overlap adder is provided to perform a time domain aliasing cancellation function, which simultaneously provides a staggered fade effect to avoid blocks. False artifacts. In general, the first decoding branch converts a single code encoded in the fourth domain to the first domain, and the second reverse processing branch executes from the third domain to the second domain a conversion, and then the converter connected to the first combiner provides a conversion from the second domain to the first domain such that the input at the combiner 600 has only the first domain signal, which is in the 3d map The decoded output signal is represented in the embodiment.

Figures 4a and 4b illustrate two different embodiments which differ in the positioning of the switch 200. In Figure 4a, the switch 200 is positioned between the output of one of the common pre-processing stages 100 and the input of the two encoded branches 400, 500. The embodiment of Fig. 4a ensures that the audio signal is only input into a single encoding branch, while another encoding branch that is not connected to the output of the common preprocessing stage is not operational and is thus turned off or in a sleep mode. This embodiment is preferred in that the inactive coding branch does not consume power and computing resources that are useful for mobile applications, particularly mobile applications that are battery powered and thus have general limitations on power consumption.

On the other hand, however, the embodiment of Figure 4b may be preferred when power consumption is not an issue. In this embodiment, the encoding branches 400, 500 are always active, and only the output of the selected encoding branch for a certain time portion and/or a certain frequency portion is forwarded to a one-bit element. The bit stream formatter implemented by stream multiplexer 800. Thus, in the embodiment of Figure 4b, both of the encoding branches are always active, and the output of an encoding branch selected by the decision stage 300 enters the output bit stream, while another unselected encoding The output of branch 400 is discarded, i.e., the output bit stream is not entered, i.e., the encoded audio signal.

Preferably, the second encoding rule/decoding rule is an LPC-based encoding algorithm. In LPC-based speech coding, a distinction is made between a quasi-periodic similar pulsed excitation signal segment or signal portion and a similar noise excitation signal segment or signal portion. This is for very much as in Figure 7b The low bit rate LPC speech coder (2.4 kbps) is executed. However, in medium rate CELP encoders, the excitation is obtained for the addition of scale vectors from an adaptive codebook and a fixed codebook.

A quasi-periodic similar to a pulsed excitation signal segment, i.e., a signal segment having a particular pitch, is encoded by a different mechanism than a similar noise-like excitation signal. When a quasi-periodic like pulse excitation signal is connected to a voiced speech, the noise-like signal is about silent speech.

For illustrative purposes, reference is made to Figures 5a through 5d. Here, quasi-periodic similar to pulse signal segments or signal portions are similarly discussed with similar noise signal segments or signal portions. Specifically, a voiced voice illustrated in the time domain of FIG. 5a and the frequency domain of FIG. 5b is discussed as an example for a quasi-periodic similar pulse signal portion, and for the 5c and 5d maps. A silent speech segment is discussed as an example of a similar noise signal portion. Speech can be generally classified as vocal, silent, or mixed. The time and frequency domain plots for the sounded and unvoiced segments for sampling are shown in Figures 5a through 5d. Voiced speech is quasi-periodic in the time domain and harmonically constructed in the frequency domain, while silent speech is similarly random and broadband. The short time spectrum of voiced speech is characterized by its fine harmonic formant structure. This fine harmonic structure is the result of quasi-periodicity of speech and contributes to the vocal chord. This formant structure (spectral envelope) is due to the interaction of the source and the vocal tract. The channel consists of the pharynx and the mouth. Due to the glottal pulse, the shape of the spectral envelope of the short time spectrum "suitable" for voiced speech is associated with the transfer characteristics of the vocal cord and spectral tilt (6 db/octave). The spectral envelope is characterized by a set of peaks called formants. These formants are these resonant modes of the vocal cords. For a typical vocal cord, there are three to five formants below 5 kHz. The amplitude and position of the first three formants (usually below 3 kHz) are important in speech synthesis level perception. It is also important to represent higher formants for wideband and unvoiced speech. These attributes of speech are relevant The following physical speech production system. Voiced speech is produced by exciting the channel with a quasi-periodic glottal air pulse generated by the vibrating vocal cord. The frequency of these periodic pulses is called the fundamental frequency or pitch. Silent speech is produced by forcing air through one of the channels to compress. The nasal sound is produced by a sudden release of air pressure formed after the closure of the passage.

Therefore, one of the audio signals, like the noise portion, does not display the time domain structure of any similar pulse as illustrated in FIG. 5c and does not display the harmonic frequency domain structure as illustrated in FIG. 5d, which is, for example, The quasi-periodic similar pulse portions illustrated in Figures 5a and 5b are different. However, as will be outlined later, a difference between a similar noise portion and a quasi-periodic similar pulse portion can also be observed after an LPC for the excitation signal. The LPC is a method of simulating a channel and extracting the excitation of the channels from the signal.

In addition, quasi-periodic-like pulse portions and similar noise portions may appear in time, that is, meaning that one portion of the audio signal is noise in time and another portion of the audio signal is quasi-periodic in time. , that is, the tone. Alternatively or additionally, the characteristics of a signal may vary in different frequency bands. Therefore, the decision whether the audio signal is noise or pitch can also be frequency selective such that a certain frequency band or bands are considered to be noise and other bands are considered to be tones. In this case, a certain time portion of the audio signal may include a tonal component and a noise component.

Figure 7a illustrates a linear model of a speech production system. This system assumes a two-level excitation, i.e., a pulse sequence as shown in Figure 7c for voiced speech and a random noise as shown in Figure 7d for silent speech. The channel is modeled as one of the pulsed all-pole filters 70 of the 7c or 7d map generated by the glottal model 72. Therefore, the system of Figure 7a can be reduced to an all-pole filter with a gain stage of Figure 7b, a transfer The path, a feedback path 79, and an add stage 80. In the feedback path 79, there is a prediction filter 81, and the following z-domain function can be used to represent the entire analog source synthesis system illustrated in Figure 7b: S(z) = g / (1-A(z) ). X(z), where g represents the gain, A(z) is the prediction filter determined by an LP analysis, X(z) is the excitation signal, and S(z) is the synthesized speech output.

Figures 7c and 7d show a graphical time domain description of one of the vocal and unvoiced speech synthesis using the linear source system model. The system and the excitation parameters in the above equation are unknown and must be determined based on a limited set of speech samples. The coefficients of A(z) are obtained using linear prediction of one of the input signals and quantization of one of the filter coefficients. In a p-th order transfer linear predictor, the current samples of the speech sequence are predicted based on a linear combination of p through sampling. The predictor coefficients can be determined by conventional algorithms, such as a Levinson-Durbin algorithm or generally an autocorrelation method or a reflection method.

Figure 7e illustrates a more detailed implementation of one of the LPC analysis blocks 510. The audio signal is input to a filter decision block that determines the filter information (A(z)). This information is output as short-term prediction information required by a decoder. The actual prediction filter 85 requires the short-term prediction information. In a subtractor 86, one of the audio signals is currently sampled and subtracted for one of the current samples to cause the line 84 to generate the prediction error signal. A sequence of such prediction error signal samples is schematically illustrated in Figure 7c or Figure 7d. Therefore, the 7a, 7b diagram can be considered as a modified similar pulse signal.

Figure 7e illustrates a preferred manner of calculating the excitation signal, and Figure 7f illustrates a preferred manner of calculating the weighted signal. In contrast to Fig. 7e, when γ is not 1, the filter 85 is different. For γ , an A value of less than 1 is preferred. Further, the block 87 appears, and μ is preferably a number less than one. In general, the elements in Figures 7e and 7f can be implemented as in 3GPP TS 26.190 or 3GPP TS 26.290.

Figure 7g illustrates one of the inverse processing that can be applied to the decoder side (such as element 537 in Figure 2b). Specifically, block 88 generates an unweighted signal from the weighted signal and block 89 calculates an excitation based on the unweighted signal. In general, all signals other than the unweighted signal in the 7g graph are processed in the LPC domain, but the excitation signal is a different signal in the same domain as the weighted signal. Block 89 outputs an excitation signal, which can then be used in conjunction with the output of block 536. This common reverse LPC conversion can then be performed at block 540 in Figure 2b.

Subsequently, a comprehensive analysis of the CELP encoder will be discussed for Figure 6 to illustrate a number of modifications to this algorithm. This CELP encoder is discussed in detail in "Speech Coding: A Tutorial Review" by Andreas Spaniasdi, IEEE Transactions 82, No. 10, pp. 1541 to 1585, October 1994. The CELP encoder illustrated in FIG. 6 includes a long-term prediction component 60 and a short-term prediction component 62. Also, use one of the code books at 64. A perceptual weighting filter W(z) is implemented at 66, and an error minimization controller is provided at 68. s(n) is the time domain input signal. After having been perceptually weighted, the weighted signal is input to a subtractor 69 to calculate an error between the weighted composite signal output at block 66 and the original weighted signal sw(n). In general, the short-term prediction filter coefficients A(z) are calculated by an LP analysis stage and their coefficients are quantized on Â(z) as shown in Figure 7e. The long-term prediction information AL(z) including the long-term prediction gain g and the vector quantization index (i.e., codebook reference) is calculated for the prediction error signal at the output of the LPC analysis stage (10a in Fig. 7e). These LTP parameters are pitch delay and gain. In CELP, this is usually implemented as an adaptive codebook containing past excitation signals (rather than residuals). The adaptive CB delay and gain are minimized by the mean squared weighted error (closed loop pitch search) Find.

Next, the CELP algorithm encodes the residual signal obtained using a codebook such as a Gaussian sequence after the short-term and long-term prediction. The ACELP algorithm (where "A" stands for "algebraic") has a codebook of a particular algebraic design.

A codebook can contain more or less vectors, each of which is some sample length. A gain factor g changes the size of the code vector and the code of the gain is filtered by the long term prediction synthesis filter and the short term prediction synthesis filter. The "best" code vector is selected such that the perceptual weighted mean square error at the output of the subtractor 69 is minimized. As illustrated in Figure 6, the search process is completed by a comprehensive analysis optimization.

For a particular situation, a TCX encoding may be more suitable for encoding the excitation in the LPC domain when the frame is mixed with one of the voiced voices or when the voice in the music appears. The TCX code processes the weighted signal in the frequency domain without making any assumptions about the excitation. The TCX is then more general than the CELP code and is not limited to one of the excitations or a silent source model. The TCX is still a source-oriented model code that simulates the formants of the speech signature signals using a linear prediction filter.

In AMR-WB+-encoding, one of the choices between different TCX modes and ACELP is known from the AMR-WB+ description. The difference in the TCX modes is that the lengths of the discrete Fourier transforms for different modes are different and the best mode can be selected by a comprehensive analysis method or a direct "feedforward" mode.

As discussed in connection with Figures 2a and 2b, the common pre-processing stage 100 preferably includes a joint multi-channel (surround/combined stereo) 101 and additionally a bandwidth extension stage 102. Correspondingly, the decoder includes a bandwidth extension stage 701 and a subsequent connected joint multi-channel stage 702. Preferably, in the case of the encoder, the joint multichannel stage 101 is The bandwidth extension stage 102 is previously connected, and at the decoder side, the bandwidth extension stage 701 is connected prior to the joint multi-channel stage 702 in terms of the signal processing direction. Alternatively, however, the common pre-processing stage may comprise one of the combined multi-channel stages in the absence of the subsequently extended bandwidth extension stage or in the absence of a connected joint multi-channel stage. Wide extension.

A preferred example of one of the combined multi-channel stages at the encoder ends 101a, 101b and at the decoder ends 702a and 702b is illustrated in the context of FIG. E number of original input channels are input to the downmixer 101a such that the downmixer produces K number of transmitted channels, wherein the number K is greater than or equal to one and less than or equal to E.

Preferably, the E input channels are input to one of the generation parameter information in conjunction with the multi-channel parameter analyzer 101b. This parameter information is preferably entropy encoded with, for example, a different encoding and subsequent Huffman encoding or optionally subsequent arithmetic encoding. The encoded parameter information output by block 101b is transmitted to parameter decoder 702b which may be part of item 702 in Figure 2b. The parameter decoder 702b decodes the transmitted parameter information and forwards the decoded information to the upmixer 702a. The upmixer 702a receives the K transmitted channel and generates L number of output channels, wherein the number L is greater than or equal to K and less than or equal to E.

Parameter information may include internal channel level differences, internal channel time differences, internal channel phase differences, and/or internal channel consistent measurements, as known from BCC techniques or as known or detailed in the MPEG Surround Standard. The number of transmission channels can be a single single channel for ultra low bit rate applications or can include a compatible stereo application or can include a compatible stereo signal or two channels. Typically, the E number of input channels can be five or possibly higher. Alternatively, as is known in the context of spatial audio object coding (SAOC), the E number of input channels may also be E audio objects.

In an implementation, the downmixer performs the original E loses One of the incoming channels is weighted or unweighted or homesick to one of the E input audio objects. If the audio object is used as an input channel, the joint multi-channel parameter analyzer 101b will calculate audio object parameters, such as a correlation between the audio objects preferably for each time portion and, more preferably, for each frequency band. matrix. For this purpose, the entire frequency range can be divided into at least 10 and preferably 32 or 64 bands.

Figure 9 illustrates a preferred embodiment of the implementation of the bandwidth extension stage 102 (in Figure 2a) and the corresponding bandwidth extension stage 701 (in Figure 2b). At the decoder side, the bandwidth extension block 102 preferably includes a low pass filtering block 102b, after the low pass or a portion of the reverse QMF, only one of the functions in the QMF band. A downsampler block and a high band analyzer 102a. The original audio signal input to the bandwidth extension block 102 is low pass filtered to produce the low frequency signal, which is then input to the encoding branches and/or the switch. The low pass filter has a cutoff frequency that can range from one of 3 kHz to 10 kHz. In addition, the bandwidth extension block 102 further includes a high frequency band analyzer for calculating the bandwidth extension parameters, such as a spectral envelope parameter information, a noise layer parameter information, and an inverse filtering parameter. Information, further parameter information about certain harmonic lines in the high frequency band and additional parameters as discussed in detail in the section on band replication in the MPEG-4 standard.

At the decoder end, the bandwidth extension block 701 includes a patch 701a, an adjustment period 701b, and a combiner 701c. The combiner 701c combines the decoded low frequency signal with the reconstructed and adjusted high frequency signal output by the adjuster 701b. A splicer provides an input to the adjuster 701b that is operative to retrieve the high frequency signal from the low frequency signal, such as through band replication or generally through bandwidth spreading. The supplementary section executed by the splicing machine may be a harmonic mode or a non-harmonic A supplementary section of the way to perform. The signal generated by the repeater 701a is then adjusted by the adjuster 701b using the transmitted parameter bandwidth extension information.

As shown in Figures 8 and 9, in the preferred embodiment, the blocks described may have a mode control input. This mode control input is derived from the decision stage 300 output signal. In this preferred embodiment, one of the characteristics of a corresponding block may be adapted to the decision level output, i.e., in one preferred embodiment, one of the speech decisions or one of the music decisions is for the audio. Made at a certain time part of the signal. Preferably, the mode control has only one or more of the functions of the blocks, and not all functions related to the blocks. For example, the decision may only affect the splicing machine 701a without affecting other blocks in FIG. 9, or may, for example, affect only the joint multi-channel parameter analyzer 101b in FIG. 8 without the eighth picture Other blocks. This implementation is preferably such that a higher flexibility and higher quality and lower bit rate output signal is obtained by being flexibly provided in the common pre-processing stage. However, on the other hand, the use of algorithms for these two signals in this common pre-processing stage allows for an efficient encoding/decoding scheme to be implemented.

Figures 10a and 10b illustrate two different implementations of the decision stage 300. An open loop decision is indicated in Figure 10a. Here, the signal analyzer 300a in the decision stage has certain rules to determine whether a particular time portion or a certain frequency portion of the input signal has a portion of the signal required by the first encoding branch 400 or the second encoding. Branch 500 is used to encode one of the characteristics. For this purpose, the signal analyzer 300a may analyze the audio input signal of the common pre-processing stage or may analyze the audio signal output by the common pre-processing stage (ie, the audio intermediate signal) or may analyze the common pre-preparation An intermediate signal in the processing stage, such as may be a single channel signal or may be the output of a downmix signal of a signal having a k channel (shown in Figure 8). At the output, the signal analyzer 300a generates the switch 200 for controlling the encoder end and the corresponding one on the decoder side. Switching decision of switch 600 or the combiner 600.

Although not discussed in detail for the second switch 521, it is emphasized that the second switch 521 can be positioned in a manner similar to the first switch 200 as discussed for Figures 4a and 4b. Thus, in Figure 3c one of the switches 521 is selectable at the output of the two processing branches 522, 523, 524 such that the two processing branches operate in parallel and only one of the processing branches outputs via the third cc One bit stream former is described as being written to a bit stream.

Additionally, the second combiner 600 can have a particular interleaved fade function as discussed in Figure 4c. Alternatively or additionally, the first combiner 532 may have the same staggered fade out function. In addition, the two combiners may have the same staggered fade-out function or may have different interlace fade-out functions or may have no interlace fade-out function at all to allow the two combiners to switch without any additional interlace fade-out functionality.

As discussed above, the two switches can be controlled by an open loop decision or a closed loop decision as discussed for Figures 10a and 10b, wherein the controllers 300, 525 in Figure 3c can have Different or the same function.

In addition, a time warping function of signal adaptation may exist not only in the first encoding branch or the first decoding branch but also on the encoder side and the second encoding branch on the decoder. The second processing branch. Depending on the processed signal, the two time warping functions may have the same time warping information such that the same time warping is applied to the signals in the first domain and the second domain. This saves throughput and may be useful in some instances where the subsequent block has a similar time warp time characteristic. However, in an alternative embodiment, it is preferred to have an independent time warping estimator for the first encoding branch and the second processing branch in the second encoding branch.

The encoded audio signal of the present invention may be stored on a digital storage medium or may be transmitted on a transmission medium such as a wireless transmission medium or a wired transmission medium such as the Internet.

In a different embodiment, the switch 200 of the 1a or 2a diagram is switched between the two encoding branches 400, 500. In a further embodiment, there may be additional coding branches, such as a third coding branch or even a fourth coding branch or even more coding branches. At the decoder end, the switch 600 of FIGURE 1b or 2b switches between the two encoding branches 431, 440 and 531, 532, 533, 534, 540. In a further embodiment, there may be additional decoding branches, such as a third decoding branch or even a fourth decoding branch or even more decoding branches. Similarly, when such additional encoding/decoding legs are provided, other switches 521 or 532 can switch between more than two different encoding algorithms.

Figure 12A illustrates a preferred embodiment of an encoder implementation, and Figure 12B illustrates a preferred embodiment of the corresponding decoder implementation. This embodiment of Figure 12A illustrates a separate sensing module 1200, and additionally illustrates the further encoder tool illustrated in block 11A, block 421, except for those elements previously discussed with corresponding parameter numbers. A preferred implementation. These additional tools are a temporal noise shaping (TNS) tool 1201 and a mid/side encoding tool (M/S) 1202. In addition, the additional functions of elements 421 and 524 are illustrated in blocks 421/542 as an implementation of a combination of scaling, noise fill analysis, quantization, and arithmetic coding of spectral values.

In the corresponding decoder implementation map 12B, additional components are illustrated, which are an M/S decoding tool 1203 and a TNS decoder tool 1204. Further, at 1205, one of the bass back filters is not illustrated in the previous figures. The transition window block 532 corresponds to the element 532 in FIG. 2B, the element 532 is illustrated as a switch but performing some may be one The transition samples are staggered out or one of the key samples is staggered and faded out. The latter is implemented as an MDCT operation in which the two-time aliasing portions are overlapped and added. Since the total bit rate can be reduced without any loss of quality, this critical sampling transition process is preferably used in the appropriate case. The additional over-windowing block 600 corresponds to the combiner 600 in FIG. 2B, which is also illustrated as a switch, but it is clear that when a zone has been processed in the first leg When the block and another block have been processed in the second minute, the component performs some sort of staggered out (critically sampled or non-critically sampled) to avoid block artifacts and specifically switch artifacts. However, when the process in the two branches perfectly matches the other, the interleaving operation can be "downgraded" to a hard handoff (and a staggered fade out operation is understood as a "between the two paths" Soft "switch".

The concepts in Figures 12A and 12B allow for the encoding of signals that are arbitrarily mixed with one of speech and audio content, and the concept performs an optimal encoding technique that is comparable or better than cropping that may be specific to speech or general audio content. The general structure of the encoder and decoder can be described as the presence of one of the MPEG Surround (MPEGS) functional units for processing stereo or multi-channel processing and one of the parameter representations for processing the higher audio frequencies in the input signal. Enhanced SBR (eSBR) One of the components is jointly pre-post-processed. Then, there are two branches, one route-modified high-order audio coding (AAC) tool path component and another route-based on linear predictive coding (LP or LPC domain) path, which is followed by the LPC residual frequency. A domain representation or a time domain is represented as a feature. The spectrum of all transmissions for AAC and LPC is represented in the MDCT domain after quantization and arithmetic coding. This time domain representation uses an ACELP excitation coding scheme. This basic structure is shown in Figure 12A for the encoder and in Figure 12B for the decoder. The data flow in this diagram is from left to right, from top to bottom. The function of the decoder is to find a description of the quantized audio spectrum or time domain representation in the bit stream payload and the quantized values and Other reconstruction information decoding.

In the case of transmitting spectral information, the decoder will reconstruct the quantized spectrum, and the reconstructed spectrum will be processed by any means active in the bit stream payload to obtain the description as described in the input bit stream payload. The actual signal spectrum, and ultimately the frequency domain is converted to the time domain. After this initial reconstruction and the scaling of the spectral reconstruction, there is an optimal tool to improve one or more spectra to provide more efficient coding.

In the case of a transmitted time domain signal representation, the decoder will reconstruct the quantized time signal and process the reconstructed time signal through any tool active in the bit stream payload to obtain the input bit stream. The actual time domain signal described by the payload.

For each of the tools operating on the signal material, the selection of "pass" is retained, and in all cases where the process is omitted, the spectrum or time sample input at it passes directly through the tool without modification.

In the bit stream from the time domain to the spectral representation or from the LP domain to the non-LP domain or vice versa, the position of the signal representation is changed, and the decoder will assist by an appropriate transition overlap-additional windowing method. The transition from one domain to another.

After the transition processing, the eSBR and MPEGS processing is applied to the two encoding paths in the same manner.

The input to the bit stream multiplexer tool is a one-way payload. The multiplexer separates the bit stream payload into portions for each tool and provides each of the tools with bit stream payload information about the tool.

The output of the bit stream payload multiplexer tool is:

●Depending on the core encoding type in the current frame, it is:

• The quantized and noise-free encoded spectrum, which is represented as follows:

●Scale factor information

●Arithmetic coding spectrum line

• Or: Linear prediction (LP) parameters and an excitation signal, which is represented by one of the following:

●Quantitative or arithmetically encoded spectral lines (conversion coding excitation, TCX) or

● ACELP coding time domain excitation

●The spectrum noise filling information (best)

● The M/S decision information (best)

●The time domain noise trimming (TNS) (best)

●The filter bank control information

● This time does not distort (TW) control information (best)

● Enhanced Band Replication (eSBR) control information

●The MPEG Surround (MPEGS) control information

The scale factor no-noise decoding tool loads the multiplexer from the bit stream to retrieve information, analyze the information, and decode the Huffman and DPCM coding scale factors.

The input to the scale factor no noise decoding tool is:

● Scale factor information for the noise-free coded spectrum

The output of the scale factor no noise decoding tool is:

● The decoded integer representation of the scale factor:

The spectrum noise-free decoding tool loads the multiplexer from the bit stream to retrieve information, analyze the information, decode the arithmetically encoded data, and reconstruct the quantized spectrum. The input to the no noise decoding tool is:

●The noise-free coding spectrum

The output of this noise-free decoding tool is:

The quantized values of the spectrum

The inverse quantizer tool extracts the quantized values for the spectrum and converts the integer values into a non-proportional adjusted, reconstructed spectrum. The quantizer is an overextension quantizer whose roll-out factor depends on the selected core coding mode.

The input to the inverse quantizer tool is:

• the quantized values for the spectrum

The output of the inverse quantizer tool is:

● The unscaled, inverse quantized spectrum

The noise fill tool is used to fill the spectral gaps in the decoded spectrum, which occur when the spectral values are quantized to zero, for example due to a very strong limit on the encoder bit requirements. The use of this noise filling tool is optimal.

The input to the noise fill tool is:

●The unscaled, inverse quantized spectrum

● Noise filling parameters

● The decoded integer representation of the scale factor

The output to the noise filling tool is:

● The unscaled, anti-spectral for the spectral line was previously quantized to zero

Vectorized spectral values.

● Improved integer representation of the scale factors

The rescaling tool converts the integer representation of the scale factors to the actual values, and multiplies the unscaled inverse quantized spectrum by the correlation scale factors.

The inputs to these scale factors are:

● The decoded integer representation of the scale factors

●The unscaled, inverse quantized spectrum

The output of the scale factor tool is:

●The unscaled, inverse quantized spectrum

For an overview of the M/S tool, please refer to ISO/IEC 14496-3, sub-clause 4.1.1.2.

For an overview of the Time Domain Noise Correction (TNS) tool, please refer to ISO/IEC 14496-3, subclause 4.1.1.2.

The filter bank/block switching tool implements the inverse of the frequency map performed in the encoder. A reverse modified discrete cosine transform (IMDCT) is used for this filter bank. The IMDCT can be configured to support 120, 128, 240, 256, 320, 480, 512, 576, 960, 1024 or 1152 spectral coefficients.

The input to the filter bank tool is:

●This (inverse quantized) spectrum

●The filter bank control information

The output of this filter tool is:

● The (equal) time domain reconstruction audio signal

The time warp filter bank/block switching tool replaces the normal filter/block switching tool when the time domain distortion mode is enabled. The filter bank is identical to the normal filter bank (IMDCT), and additionally, the windowed time domain samples are mapped from the warped time domain to the linear time domain by time varying resampling.

The inputs to these time warp filter bank tools are:

●The inverse quantized spectrum

●The filter bank control information

●The time warp control information

The output of this filter bank tool is:

● The (equal) linear time domain reconstruction audio signal

The enhanced SBR (eSBR) tool reproduces the high frequency of the audio signal. It is a copy of these sequences based on harmonics that are truncated during encoding. It adjusts the resulting high frequency spectral envelope and applies inverse filtering, and adds noise and sinusoidal components to reproduce the spectral characteristics of the original signal.

The input to the eSBR is:

●Quantitative envelope data

●Miscellaneous control data

• a time domain signal from the AAC core decoder

The output of the eSBR is:

● a time domain signal or

a QMP field representation of a signal, such as the use of the MPEG ring

In the case of a tool.

The MPEG Surround (MPEGS) can generate a plurality of signals from one or more input signals by applying a complex upmix procedure to the (iso) input signal controlled by appropriate spatial parameters. In the USAC context, MPEGS uses a transmission parameter side-by-side information and a transmission downmix signal to encode a multi-channel signal.

The input to the MPEGS tool is:

● a downmix time domain signal or

• A QMF domain representation of a downmix signal from the eSBR tool

The output of the MPEGS tool is:

● A multi-channel time domain signal

The signal classifier tool analyzes the original input signal and thereby generates control information that triggers selection of the different coding modes. The input The analysis of the signal is dependent on the implementation and will attempt to select the best core coding mode for a given input signal frame. The output of the signal classifier can also be used (optimally) to influence the performance of other tools, such as MPEG Surround, Enhanced SBR, Time Warped Filter Banks, and others.

The input to the Signal Classifier tool is:

●The original unimproved input signal

● Additional implementation of dependent parameters

The output of this signal classifier tool is:

Controlling the selection of one of the core codecs (non-LP filter frequency domain coding, LP filtering frequency domain or LP filtering time domain coding)

In accordance with the present invention, the time/frequency resolution in block 12A and in converter 523 in Fig. 12A is controlled in dependence on the audio signal. The relationship between the window length, the conversion length, the temporal resolution, and the frequency resolution is explained in Fig. 13A, wherein it becomes clear that for a long window length, the temporal resolution becomes low but the frequency resolution becomes High, and for a short window length, the time resolution becomes higher but the frequency resolution becomes lower.

In the first encoding branch (preferably the AAC encoding branch indicated by elements 410, 1201, 1202, 4021 of Figure 12A), different windows may be used, wherein the window shape is comprised by a signal analyzer It is decided that the signal analyzer is encoded in the signal classification block 300 but it can also be a separate module. The encoder selects a window among the windows having different time/frequency resolutions as illustrated in FIG. 13B. The time/frequency resolution of the first long window, the second long window, the third long window, the fourth long window, the fifth long window, and the sixth long window is equal to 2048 sample values (for 1024 One conversion length). Said in the third line in Figure 13B The short window of Ming has a temporal resolution corresponding to 256 sample values of its window size. This corresponds to a conversion length of 128.

Similarly, the last two windows have a window length equal to 2304, which has a better frequency resolution and a lower temporal resolution than the window in the first line. The conversion length of the windows in the last two lines is equal to 1152.

In the first coding branch, different window sequences can be constructed based on the conversion windows in Figure 13B. Although only a short sequence is illustrated in Fig. 13C, while other "sequences" consist of only a single window, a larger sequence of multiple windows may be constructed. Note that, according to Figure 13B, for a smaller number of coefficients, i.e., 960 instead of 1024, the temporal resolution is also less than the corresponding higher number of coefficients, such as 1024.

14A through 14G illustrate different resolution/view sizes in the second encoding branch. In a preferred embodiment of the invention, the second encoding branch has a first processing branch (which is an ACELP time domain encoder 526), and the second processing branch includes the filter bank 523. In this branch, a hyperframe, such as a 2048 sample, is subdivided into 256 sample frames. The individual frames of 256 samples can be used separately so that one of the four windows (each window covers two frames) can be applied when the application has one of the MDCTs with 50 percent overlap. Next, as explained in Fig. 14D, a high temporal resolution is used. Alternatively, when the signal allows for a longer window, the sequence as in Figure 14C can be applied, where the application has one double window size of 1024 samples for each window (medium window) so that a window is covered Four frames and there is an overlap of 50%.

Finally, when the signal is such that a long window is used, the long window expands by 4096 samples and also has a 50% overlap.

There are two branches (one of which has an ACELP series) In the preferred embodiment of the encoder, the position of the ACELP frame indicated by "A" in the superframe may also determine the application of two adjacent TCX frames indicated by "T" in FIG. 14E. The size of the window. Basically, people are interested in using long windows as much as possible. However, when a single T frame is between two A frames, a short window must be applied. A medium window is applied when there are two adjacent T frames. However, when there are three adjacent T frames, a corresponding larger window may not be efficient due to additional complexity. Therefore, although the third T frame is not in front of an A frame, it can be processed by a short window. A long window is applied when the entire hyperframe has only T frames.

Figure 14F illustrates several options for a window where the window size is always 2x of the number of spectral coefficients 1g due to a better 50% overlap. However, other overlap percentages for all coded branches may be applied such that the relationship between window size and transition length may also differ from two and even close to one when no time domain aliasing is applied.

Figure 14G illustrates the rules for constructing a window based on the rules given in Figure 14F. This value ZL indicates the zero at the beginning of the window. This value L illustrates a plurality of window coefficients in an aliasing region. The equivalent in section M is a "1" value that does not introduce any aliasing, since the portion corresponding to M overlaps with one of an adjacent window having a value of zero. This portion M is followed by a right overlap region R, which is followed by a ZR region of zero, which will correspond to a portion M of a subsequent window.

Reference is made to the appended annex, which describes a preferred and detailed implementation of an inventive audio encoding/decoding scheme, particularly with respect to the decoder side.

annex 1. Window and sequence

Quantization and coding are done in this frequency domain. For this purpose, in the encoder, the time signal is mapped into the frequency domain. The decoder performs as in striping The reverse mapping in paragraph 2. Depending on the signal, the encoder can change the time/frequency resolution by using three different window sizes: 2304, 2048 and 256. In order to switch between windows, use these transition windows LONG_START_WINDOW, LONG_STOP_WINDOW, START_WINDOW_LPD, STOP_WINDOW_1152, STOP_START_WINDOW and STOP_START_WINDOW_1152. Table 5.11 lists the windows, specifies the corresponding transition lengths and schematically shows the shape of the windows. Use three conversion lengths: 1152, 1024 (or 960) (reference long conversion) and 128 (or 120) coefficients (reference short conversion).

The window sequence consists of a window in which one raw_data_block always contains one of the data representing 1024 (or 960) output samples. The data element window_sequence indicates the window sequence actually used. Figure 13C shows how these window sequences are composed of individual windows. Refer to Sub-Clause 2 for more detailed information on the conversion and the windows.

1.2 Scale factor band and grouping

See ISO/IEC 14496-3, subpart 4, subclause 4.5.2.3.4

As explained in ISO/IEC 14496-3, subpart 4, subclause 4.5.2.3.4, the width of the scale factor bands is based on the imitation of these critical bands of the human auditory system. For this reason, the number of scale factor bands in a spectrum and their width depend on the conversion length and the sampling frequency. Tables 4.110 to 4.128 in Section 4.4.5 of ISO/IEC 14496-3 subsection 4 list the frequency bands for each of these conversion frequencies on the conversion lengths 1024 (960) and 128 (120). The offset of the beginning. original These tables designed for LONG_WINDOW, LONG_START_WINDOW and LONG_STOP_WINDOW are also available for START_WINDOW_LPD and STOP_START_WINDOW. Tables 4 through 10 are the offset tables for STOP_WINDOW_1152 and STOP_START_WINDOW_1152.

1.3 decoding of lpd_channel_stream()

The lpd_channel_stream() bitstream element contains all the necessary information to decode one of the "linear prediction domain" encoded signals. It contains a payload for a frame of coded signals encoded in the LPC domain (i.e., including an LPC filtering step). The residual of this filter (so called "excitation") is then represented by the help of an ACELP module or in the MDCT conversion domain ("conversion coding excitation", TCX). In order to allow close adaptation to these signal characteristics, a frame is divided into four smaller units of equal size, each smaller unit being encoded with an ACELP or TCX coding scheme.

This process is similar to the coding scheme described in 3GPP TS 26.290. Inheriting this file is a slightly different term, where a "superframe" represents one of the 1024 samples of a signal segment, and a "frame" is exactly one quarter of the segment of the signal, ie 256 samples. Each frame in these frames is further subdivided into four sub-frames of equal length. Please note that this sub-chapter uses this term.

1.4 definition, data components

Acelp_core_mode In the case where ACELP is used as an lpd coding mode, this bit field indicates an accurate bit allocation scheme.

Lpd_mode This bit field mode defines these encoding modes for each of the four frames in a hyperframe (corresponding to an AAC frame) in lpd_channel_stream(). The encoding modes are stored in the array mod[] and are taken from 0 to 3. Table 1 below can determine the mapping from lpd_mode to mod[].

The value of mod[0..3] in the array mod[] indicates the respective encoding mode in each frame:

Acelp_coding() contains the syntax elements for all data decoded by the ACELP excitation frame.

Tcx_coding() contains syntax elements for all data decoded based on MDCT Transformation Code Excitation (TCX).

First_tcx_flag indicates whether the currently processed TCX frame is the flag of the first frame in the superframe.

Lpc_data() contains the syntax elements that decode the decoding of all LPC filter parameters required to decode the current frame.

First_lpd_flag indicates whether the current frame is the first frame in a sequence of hyperframes encoded in the LPC domain. According to Table 3, the bit stream element core_mode (in the case of channel_pair_element is The history of core_mode0 and core_mode1) can also determine this flag.

Last_lpd_mode indicates the lpd_mode of the previously decoded frame.

1.5 decoding process

The order of decoding in the lpd_channel_stream is:

Get acelp_core_mode

Get lpd_mode and decide the content of the auxiliary variable mod[]

Get acelp_coding or tcx_coding data, depending on auxiliary variables

Depending on the content of mod[]

Get lpc_data

1.6 ACELP/TCX coding mode combination

Similar to [8] in Section 5.2.2, there is a combination of 26 allowed ACELP or TCX in one of the lpd_channel_stream payloads. Each of these 26 mode combinations is marked in the bit stream element lpd_mode. The mapping of lpd_mode to the actual coding mode for each frame in a subframe is shown in Table 1 and Table 2.

1.7 Scale Factor Band Table Reference For all other scale factor band tables, please refer to ISO/IEC 14496-3 subsection 4, Section 4.5.4, Table 4.129 to Table 4.147.

1.8 Quantization To quantify the AAC spectral coefficients in the encoder, a non-uniform quantizer is used. Therefore, the decoder must perform reverse non-uniform quantization after the equal scale factor Huffman decoding (see subclause 6.3) and no noise decoding of the spectral data (see subclause 6.1).

To quantify these TCX spectral coefficients, a uniform quantizer is used. The decoder does not require inverse quantization after noise-free decoding of the spectral data.

2. Filter bank and block switching 2.1 Tool Description

The time/frequency representation of the signal is mapped to the time domain by feeding it to the filter module. The module consists of a reverse modified discrete cosine transform (IMDCT) and a window and an overlap function addition function. In order to adapt the time/frequency resolution of the filter bank to the characteristics of the input signal, a block switching tool is also used. N represents the length of the window where N is a function of the window_sequence (see subclause 1.1). For each channel, the N/2 time-frequency value is converted to the N time domain value x _i,n through IMDCT. After applying the window function, for each channel, the first half of the z _i,n sequence is added to the previous block windowing sequence z( _{i-1), and} the second half of _n is reconstructed for each channel These output samples of out _i,n .

2.2 Definition

Window_sequence indicates which window sequence (ie block size) is used for 2 bits.

The window_shape indicates which one of the window functions is selected.

Figure 13C shows eight window_sequences (ONLY_LONG_SEQUENCE, LONG_START_SEQUENCE, EIGHT_SHORT_SEQUENCE, LONG_STOP_SEQUENCE, STOP_START_SEQUENCE, STOP_1152_SEQUENCE, LPD_START_SEQUENCE, STOP_START_1152_SEQUENCE).

In the following, LPD_SEQUENCE refers to all allowed window/encoding mode combinations in the so-called linear prediction domain codec (see Section 1.3). In the context of decoding a frequency domain coded frame, it is important to know that only one subsequent frame is encoded with the LP domain coding mode represented by an LPD_SEQUENCE. However, when decoding the LP domain coded frame, the exact structure in the LPD_SEQUENCE is noted.

2.3 decoding process 2.3.1 IMDCT

The analytical form of the IMDCT is:

Where: n = sampling index

i=window index

k=Spectrum coefficient index

N = window length based on the window_sequence value

n ₀ = (N/2 + 1) / 2

The analysis window length N for the reverse conversion is a function of the syntax element window_sequence and the context of the algorithm. Its definition is as follows: window length 2304:

Window length 2408:

These major block transitions are as follows: From ONLY_LONG_SEQUENCE to

From LONG_START_SEQUENCE to

From LONG_STOP_SEQUENCE to

From EIGHT_SHORT_SEQUENCE to

From LPD_SEQUENCE to

From STOP_START_SEQUENCE to

From LPD_START_SEQUENCE to

From STOP_1152_SEQUENCE to

From STOP_START_1152_SEQUENCE to

2.3.2 Windowing and Block Switching

Depending on the window_sequence and window_shape components, different conversion windows are used. One of a plurality of half windows as described below provides all possible window_sequences.

For window_shape ==1, these window coefficients are given by Kaiser-Bessel derived (KBD) window as follows: Where: w ', Caesar Besso kernel window function (see also [5]) is defined as follows: α = kernel window alpha factor, Otherwise, for window_shape ==0, use a sine function as follows:

The window length N for the KBD and the sinusoidal window may be 2048 (1920) or 256 (240). In the case of STOP_1152_SEQUENCE and STOP_START_1152_SEQUENCE, N can still be 2048 or 256, and the window tilts are similar, but the flat top region is longer.

Only in the case of LPD_START_SEQUENCE, the right part of the window is a sine window of 64 samples.

Sections a)-h) of this subclause explain how to obtain such possible window sequences.

For all kinds of window_sequences, the window_shape of the left half of the first conversion window is determined by the shape of the window of the previous block. The following formula indicates this fact:

Wherein: window_shape_previous_block: a front block (i-1) of the window_shape. For the first raw_data_block() to be decoded, the window_shape of the left and right half of the window is the same.

a) ONLY_LONG_SEQUENCE:

The window_sequence == ONLY_LONG_SEQUENCE equal to 2048 (1920) of a total window length LONG_WINDOW of N_l.

For window_shape ==1, this window for ONLY_LONG_SEQUENCE is given as follows:

If window_shape ==0 for ONLY_LONG_SEQUENCE this window can be described as follows:

After windowing, the time domain values (zi, n) can be expressed as; z _i,n = w ( n ). x _i,n ;

b) LONG_START_SEQUENCE:

The LONG_START_SEQUENCE is needed to obtain a correct overlap and to add a block transition from an ONLY_LONG_SEQUENCE to an EIGHT_SHORT_SEQUENCE.

N_l window length and N_s are set to 2048 (1920) and 256 (240).

If window_shape ==1, the window for LONG_START_SEQUENCE can be given as follows: If window_shape==0, the window for LONG_START_SEQUENCE looks like: The windowed time domain value can be calculated using the formula described in a).

c) EIGHT_SHORT

The window_sequence ==EIGHT_SHORT contains eight overlapping and joined SHORT_WINDOW, each SHORT_WINDOW has a length N_s of 256 (240). The total length of the window_sequence and the leading and trailing zeros are 2048 (1920). Each of the eight blocks is first individually windowed. By the variable j = 0, ..., M -1 (M = N_l / N_s) as an index to the number of the segment block.

The window_shape of the previous block only affects the first short block in the eight short block (W0(n)). If window_shape ==1, these window functions can be given as follows:

Otherwise, if window_shape ==0, these window functions can be described as:

The overlap and addition between the EIGHT_SHORT window_sequences (the resulting windowed time domain values zi, n) are described as follows:

d) LONG_STOP_SEQUENCE

This window_sequence needs to be switched back from an EIGHT_SHORT_SEQUENCE to an ONLY_LONG_SEQUENCE.

If window_shape ==1, the window for LONG_STOP_SEQUENCE is given as follows:

If window_shape==0, the window for LONG_START_SEQUENCE is determined as follows:

The windowed time domain values can be calculated using the formula described in a).

e) STOP_START_SEQUENCE:

When only one ONLY_LONG_SEQUENCE is needed, the STOP_START_SEQUENCE is required for a block transition from an EIGHT_SHORT_SEQUENCE to an EIGHT_SHORT_SEQUENCE to obtain a correct overlap and addition.

N_l window length and N_s are set to 2048 (1920) and 256 (240).

If window_shape==1, the window for STOP_START_SEQUENCE is given as follows: If window_shape==0, the window for STOP_START_SEQUENCE looks like:

The windowed time domain values can be calculated using the formula described in a).

f) STOP_START_SEQUENCE:

The LPD_SEQUENCE is required for a block transition from an ONLY_LONG_SEQUENCE to an LPD_SEQUENCE to obtain a correct overlap and addition.

N_l window length and N_s are set to 2048 (1920) and 256 (240).

If window_shape ==1, the window for LPD_START_SEQUENCE is given as follows: If window_shape ==0, the window for LPD_START_SEQUENCE looks like: The windowed time domain values can be calculated using the formula described in a).

g) STOP_1152_SEQUENCE:

The STOP_1152_SEQUENCE is required for a block transition from an LPD_SEQUENCE to an ONLY_LONG_SEQUENCE to obtain a correct overlap and addition.

N_l window length and N_s are set to 2048 (1920) and 256 (240).

If window_shape ==1, the window for STOP_1152_SEQUENCE is given as follows:

If window_shape ==0, the window for STOP_1152_SEQUENCE is given as follows:

The windowed time domain values can be calculated using the formula described in a).

h) STOP_START_1152_SEQUENCE:

When only one ONLY_LONG_SEQUENCE is needed, the STOP_START_1152_SEQUENCE is required for a block transition from an LPD_SEQUENCE to an EIGHT_SHORT_SEQUENCE to obtain a correct overlap and addition.

N_l window length and N_s are set to 2048 (1920) and 256 (240).

If window_shape ==1, the window for STOP_START_SEQUENCE is given as follows: If window_shape==0, the window for STOP_START_SEQUENCE looks like: The windowed time domain values can be calculated using the formula described in a).

2.3.3 overlap and addition to the previous window sequence

In addition to the overlap and addition in the EIGHT_SHORT window_sequence , the first (left) portion of each window_sequence overlaps and adds to the second (right) portion of the previous window_sequence to produce a final time domain value out _i,n . The mathematical expression of this operation can be described as follows: in the case of ONLY_LONG_SEQUENCE, LONG_START_SEQUENCE, EIGHT_SHORT_SEQUENCE, LONG_STOP_SEQUENCE, STOP_START_SEQUENCE, LPD_START_SEQUENCE:

And in the case of STOP_1152_SEQUENCE STOP_START_1152_SEQUENCE:

In the case of LPD_START_SEQUENCE, the next sequence is LPD_SEQUENCE. A SIN or KBD window is applied to LPD_SEQUENCE for a good overlap and addition.

In the case of STOP_1152_SEQUENCE, STOP_START_1152_SEQUENCE, the previous sequence is LPD_SEQUENCE. A TDAC is applied to LPD_SEQUENCE to achieve a good overlap and addition.

3. IMDCT See sub-clause 2.3.1 3.1 Windowing and block switching

Depending on the window_shape component, a different oversampling conversion window prototype is used. The length of the oversampling window is: N _OS = 2. n _ long. Os _ factor _ win

For window_shape ==1, these window coefficients are given by the Caesar Besso derived (KBD) window as follows:

Among them, W ', Caesar Besso derived kernel window function (see also [5]) is defined as follows: α = kernel window alpha factor, α = 4

Otherwise, for window_shape ==0, use a sine window as follows:

For various window_sequences, the prototype used for the left window is determined by the shape of the window of the previous block. The following formula expresses this fact: Similarly, the prototype for the shape of the right window is determined by the following formula:

Since the transition length has been determined, only the difference between EIGHT_SHORT_SEQUENCE and all others must be indicated: a) EIGHT SHORT SEQUENCE: The following section similar to c-code describes the windowing and internal overlap-addition of an EIGHT_SHORT_SEQUENCE:

4. MDCT-based TCX 4.1 Tool Description

When the core_mode is equal to 1 and when one or more modes of the three TCX modes are selected as the "linear prediction domain" coding, that is, one of the 4 array items of mod[] is greater than 0, the MDCT-based is used. TCX tool. The MDCT-based TCX receives the quantized spectral coefficients from the arithmetic decoder. The quantized coefficients are first completed by a comfort noise before an inverse MDCT conversion is applied to obtain a time domain weighted synthesis (which is then fed to the weighted synthesis LPC filter).

4.2 Definition

4.3 decoding process

The MDCT-based TCX requests the arithmetic decoder for a plurality of quantized spectral coefficients lg, which are determined by the mod[] and last_lpd_mode values. These two The value also defines the length and shape of the window to be applied to the inverse MDCT. The window consists of three parts: a left end overlap of the L samples, a middle part of one of the M samples, and a right overlap of the R samples. In order to obtain an MDCT window of length 2*lg, ZL zero is added to the left end and ZR zero is added to the right end, as shown in the 14G/14F chart for Table 3.

The MDCT window is given by The quantized spectral coefficients, quant[] transmitted by the arithmetic decoder are completed by a comfort noise. The level of the injected noise is determined by the decoding noise_factor as follows: noise_level=0.0625*(8-noise_factor)

Then use a random function, random_sign(), a randomly transmitted value of -1 or +1 to compute a noise vector, noise[].

Noise[i]=random_sign()*noise_level; combining the quanti[] and noise[] in such a way that the multiple consecutive 8 consecutive zeros in quanti[] are replaced by the components in noise[] The reconstructed spectral coefficient vector r[] is formed. According to the formula to detect a series of 8 non-zero:

The reconstructed spectrum is obtained as follows:

Before performing the inverse MDCT, a spectrum de-shaping is performed according to the following steps:

1. Calculate the energy E _{m of} the 8-dimensional block at the index m for each of the first quarter of the spectrum.

2. The operation ratio R _m = sqrt ( E _m / E _I ), where I is the block index with the largest of all E _m

3. If R _m < 0.1 , set R _m = 0.1

4. If R _m < R _m-1 , set R _m = R _m-1

The factor Rm is then multiplied by each of the first quarter of the spectrum belonging to the first quarter.

The reconstructed spectrum is fed in an inverse MDCT. The non-visualized output signal x[] is scaled by a gain g obtained by inverse quantization of one of the decoded global_gain indices: g = 10 ^{global_gain / 28 / (2. rms )}

Where rms is calculated as:

Then the solution is adjusted to synthesize the time domain signal equal to: x _w [ i ]= x [ i ]. g

After the solution is scaled, the windowing and overlap addition are applied.

The reconstructed TCX target x(n) then passes through the zero state inverse weighted synthesis filter ( z )(1- αz ^-1 )/( ( z / λ ) filtering to find the synthesis filter. Note that the inserted LP filter is used for each frame in this filtering. Once the excitation is determined, the signal is filtered by the excitation filter 1/ ( z ) and then reconstructed by filtering through the filter 1/(1-0.68z ^-1 ) as described above.

Note that in a subsequent frame, the firing also requires updating the ACELP Adaptive Codebook and allowing switching from TCX to ACELP. It should also be noted that the length of the TCX synthesis is given by the TCX frame length (no overlap) for mod[] of 1, 2, 3, respectively: 256, 512 or 1024 samples.

Specification reference

[1] ISO/IEC 11172-3:1993, Information technology-Coding of moving pictures and associated audio for digital storage media at up to about 1,5 Mbit/s, Part 3: Audio.

[2] ITU-T Rec. H.222.0 (1995) | ISO/IEC 13818-1:2000, Information technology-Generic coding of moving pictures And associated audio information:-Part 1: Systems.

[3] ISO/IEC 13818-3:1998, Information technology-Generic coding of moving pictures and associated audio information:-Part 3: Audio.

[4] ISO/IEC 13818-7:2004, Information technology-Generic coding of moving pictures and associated audio information:-Part 7: Advanced Audio Coding (AAC).

[5] ISO/IEC 14496-3:2005, Information technology-Coding of audio-visual objects-Part 1: Systems

[6] ISO/IEC 14496-3:2005, Information technology-Coding of audio-visual objects-Part 3: Audio

[7] ISO/IEC 23003-1:2007, Information technology-MPEG audio technologies-Part 1: MPEG Surround

[8] 3GPP TS 26.290 V6.3.0, Extended Adaptive Multi-Rate-Wideband (AMR-WB+) codec; Transcoding functions

[9] 3GPP TS 26.190, Adaptive Multi-Rate-Wideband (AMR-WB) speech codec; Transcoding functions

[10] 3GPP TS 26.090, Adaptive Multi-Rate (AMR) speech codec; Transcoding functions

definition

The definition can be found in ISO/IEC 14496-3 subpart 1 subsection 1.3 (terms and definitions) and 3GPP TS 26.290 section 3 (definitions and abbreviations).

Although some aspects have been described in the context of a device, However, it is clear that these levels also represent one of the corresponding methods, wherein a block or device corresponds to one of the method steps or one of the method steps. Similarly, the level described in the context of a method step also represents a block or item or feature corresponding to one of the corresponding devices.

The encoded audio signal of the present invention can be stored on a digital storage medium or can be transmitted on a transmission medium such as a wireless transmission medium or a wired transmission medium such as the Internet.

Depending on certain implementation requirements, embodiments of the present invention may be implemented in hardware or software, and the implementation may be performed using a digital storage medium, such as a floppy disk, a DVD, a CD, a ROM, a PROM, An EPROM, an EEPROM or a flash memory (FLASH) memory having electrically readable control signal storage thereon for cooperating (or capable of assisting) with a programmable computer system to enable the described herein One of the methods is executed.

Some embodiments in accordance with the present invention comprise a data carrier having an electrically readable control signal that is capable of cooperating with a programmable computer system such that one of the methods described herein is carried out.

Other embodiments comprise a computer program stored on a machine readable carrier for performing one of the methods described herein.

In other words, an embodiment of the method of the present invention is thus a computer program having one of the computer programs for performing one of the methods described herein, when the computer program is executed on a computer.

A further embodiment of the method of the present invention is thus a data carrier (or a digital storage medium or a computer readable medium) comprising: recorded thereon for performing one of the methods described herein One of the methods of the computer program.

A further embodiment of one of the methods of the invention is thus a table A data stream or a sequence of signals of the computer program for performing one of the methods described herein. The data stream or the sequence of signals can be configured, for example, to be transmitted over a data communication connection (e.g., via the Internet).

A further embodiment comprises a processing device, such as a computer or a programmable logic device, configured or modified to perform one of the methods described herein.

A further embodiment includes a computer having the computer program installed thereon for performing one of the methods described herein.

In some embodiments, a programmable logic device (e.g., a field programmable gate array) can be used to perform some or all of the functions of the methods. In some embodiments, a field programmable gate array can cooperate with a microprocessor to perform one of the methods described herein. In general, the methods are preferably performed by any hardware device.

The above described embodiments are merely illustrative of the principles of the invention. It will be apparent that modifications and variations of the configurations described herein and the details are apparent to those skilled in the art. Therefore, the intention is to be limited only by the scope of the appended claims

200‧‧‧ switch

300, 525‧‧‧Signal Analyzer

400‧‧‧First Coded Branch

410‧‧‧ first converter

410a, 523a‧‧‧ windowizer

410b, 523b‧‧‧ converter

421‧‧‧Quantizer/Encoder Level

500‧‧‧Second code branch

510‧‧‧LPC processor, domain converter

523‧‧‧Second converter

524‧‧‧ Further coding tools

800‧‧‧output interface

801‧‧‧Encoder output signal

Claims (19)

An audio encoder for encoding an audio signal, comprising: a first encoding branch for encoding an audio signal using a first encoding algorithm to obtain a first encoded signal, the first encoding branch The circuit includes a first converter for converting an input signal into a spectral domain; a second encoding branch for encoding an audio signal using a second encoding algorithm to obtain a second encoded signal, wherein The first encoding algorithm is different from the second encoding algorithm, and the second encoding branch includes a domain converter for converting an input signal from an input field to an output domain and for converting an input signal into a a second converter of the spectral domain; a switch for switching between the first encoding branch and the second encoding branch such that for a portion of the audio input signal, the first encoded signal or the first The second encoded signal is in an encoder output signal; a signal analyzer is configured to analyze the portion of the audio signal to determine that the portion of the audio signal is represented as the first in the encoder output signal An encoded signal or the second encoded signal, wherein the signal analyzer is further configured to variably determine the first converter and when the first encoded signal or the second encoded signal representing the partial audio signal is generated a time/frequency resolution of the second converter; and an output interface for generating an encoder output signal, the encoder output signal comprising the first encoded signal and the second encoded signal And indicating an indication of one of the first encoded signal and the second encoded signal, and indicating one of the time/frequency resolutions for encoding the first encoded signal and for encoding the second encoded signal. The audio encoder of claim 1, wherein the signal analyzer is configured to classify the partial audio signal into an audio signal with pseudo-speech or an audio signal with pseudo-music, and In the case of a music signal, a transient detection is performed to determine the time/frequency resolution of the first converter or to perform a comprehensive analysis process to determine the time/frequency resolution of the second converter. The audio encoder of claim 1 or 2, wherein the first converter and the second converter comprise a variable windowing conversion processor, the variable windowing conversion processor comprising A window function of a variable window size and a conversion function having a variable conversion length, and wherein the signal analyzer is configured to control the window size and/or the conversion length based on the signal analysis. The audio encoder of claim 1, wherein the second encoder branch includes a first processing branch for processing an audio signal in the domain determined by the domain converter and A second processing branch including one of the second converters, wherein the signal analyzer is configured to subdivide the portion of the audio signal into a series of sub-portions, and wherein the signal analyzer is configured to The sub-portion processed by the first processing branch is determined relative to the position of a sub-portion of the portion processed by the second processing branch The time/frequency resolution of the second converter. The audio encoder of claim 4, wherein the first processing branch comprises an ACELP encoder, wherein the second processing branch comprises an MDCT-TCX processing device, wherein the signal analyzer is configured Setting the time resolution of the second converter to determine a first value from a length of one of the sub-portions or a second value from a length of one of the sub-portions that is an integer value greater than one The second value is less than the first value. The audio encoder of claim 1, wherein the signal analyzer is configured to determine a signal classification in a constant grating covering a plurality of equal-sized audio sampling blocks, and to The block is subdivided into a variable number of blocks according to the audio signal, wherein a length of one of the sub-blocks determines the first time/frequency resolution or the second time/frequency resolution. The audio encoder of claim 1, wherein the second encoding branch comprises: a first processing branch for processing an audio signal; a second processing branch, the second processing branch Including the second converter; and a further switch for switching between the first processing branch and the second processing branch such that a portion of the audio signal is input to the second encoding branch A first processed signal or a second processed signal is in the second encoded signal. A method of encoding an audio signal of an audio signal, comprising the steps of: Using a first encoding algorithm to encode an audio signal to obtain a first encoded signal in a first encoding branch, the first encoding branch including the first converter for converting an input signal into a spectral domain Converting an audio signal to a second encoded signal using a second encoding algorithm in a second encoding branch, wherein the first encoding algorithm is different from the second encoding algorithm, the second encoding branch a second converter for converting an input signal from an input field to an output field and a second converter for converting an input signal into a spectral domain; the first encoding branch and the second encoding Switching between the branches such that for a portion of the audio input signal, the first encoded signal or the second encoded signal is in an encoder output signal; analyzing the portion of the audio signal to determine the portion of the audio signal at the encoder output Is the signal represented as the first encoded signal or the second encoded signal, when the first encoded signal or the second encoded signal representing the portion of the audio signal is generated, Determining a respective time/frequency resolution of the first converter and the second converter; and generating an encoder output signal, the encoder output signal comprising the first encoded signal and the second encoded signal and Instructing one of the first encoded signal and the second encoded signal to indicate, and indicating one of the time/frequency resolutions for encoding the first encoded signal and for encoding the second encoded signal. An audio decoder for decoding an encoded signal, the decoded signal packet Having a first coded signal, a second coded signal, indicating one of the first coded signal and the second coded signal, and decoding one of the first coded signal and the second coded audio signal for time/frequency analysis Instructed by the degree, the audio decoder includes: a first decoding branch for decoding the first encoded signal using a first controllable frequency/time converter, the first controllable frequency/time converter The time/frequency resolution indication configured to be controlled to use the first encoded signal to obtain a first decoded signal; a second decoding branch for using a second controllable frequency/time converter Decoding the second encoded signal, the second controllable frequency/time converter being configured to be controlled to use the time/frequency resolution indication of the second encoded signal; a controller for using the time a frequency resolution indicator for controlling the first frequency/time converter and the second frequency/time converter; a domain converter for generating a composite signal using the second decoded signal; and a combiner, its And combining the first decoded signal with the synthesized signal to obtain a decoded audio signal. The audio decoder of claim 9, wherein the second decoding branch includes a first reverse processing branch, and the first reverse processing branch is additionally included in the reverse processing Decoding a first processed signal to obtain a first reverse processed signal; wherein the second controllable frequency/time converter is located at a second In the reverse processing branch, the second reverse processing branch is configured to reverse process the second encoded signal in the same domain as the first inverted processed signal to obtain a second reverse Processing a signal; a further combiner for combining the first inverse processed signal with the second reverse processed signal to obtain a combined signal; and wherein the combined signal is input to the combiner. The audio decoder of any one of the preceding claims, wherein the first frequency/time converter and the second frequency/time converter are time domain aliasing cancellation converters, the time domain aliasing The cancellation converter has a one-time overlap/add unit for canceling a time domain alias included in the first encoded signal and the second encoded signal. The audio decoder according to any one of the preceding claims, wherein the encoded signal includes an encoding mode indication for identifying whether an encoded signal is the first encoded signal and the second encoded signal, and wherein the decoder Further included is an input interface for interpreting the encoding mode indication to determine whether the encoded signal is to be fed to the first decoding branch or to the second decoding branch. The audio decoder of claim 9, wherein the first encoded signal is arithmetically encoded, and wherein the first encoding branch comprises an arithmetic decoder. The audio decoder of claim 9, wherein the first encoding branch comprises a dequantizer having a non-uniform dequantization characteristic for canceling when the first encoded signal is generated One of the results of a non-uniform quantization performed, Wherein the second encoding branch comprises a dequantizer using different dequantization characteristics, or wherein the second encoding branch does not comprise a dequantizer. The audio decoder of claim 9, wherein the controller is configured to apply one of a plurality of potentially different discrete frequencies/time resolutions to each converter by a discrete frequency/time resolution Controlling the first frequency/time converter and the second frequency/time converter, the number of possible different discrete frequencies/time resolutions of the second converter being higher than the possible different frequency/time resolution of the first converter The number of degrees. The audio decoder of claim 9, wherein the domain converter is an LPC synthesis processor that uses a PC filter information to generate the synthesized signal, the LPC filter information being included in the encoded signal. . A method for audio decoding an encoded signal, the encoded signal comprising a first encoded signal, a second encoded signal, an indication indicating the first encoded signal and the second encoded signal, and a decoding used to decode the first encoding a time/frequency resolution indication of the signal and the second encoded audio signal, the method comprising the steps of: decoding, by the first decoding branch, a first controllable frequency/time converter to decode the first encoded signal, the first A controllable frequency/time converter is configured to be controlled to use the time/frequency resolution indication of the first encoded signal to obtain a first decoded signal; to use a second controllable frequency by a second decoding branch /time converter to decode the second encoded signal, the second controllable frequency/time The converter is configured to be controlled to use the time/frequency resolution indication of the second encoded signal; using the time/frequency resolution indication to control the first frequency/time converter and the second frequency/time converter Generating a composite signal using the second decoded signal by a domain converter; and combining the first decoded signal with the composite signal to obtain a decoded audio signal. A computer readable medium having a coded audio signal stored thereon, comprising: a first coded signal; a second coded signal, wherein the first coded signal or the second coded signal is used to represent an audio signal a portion; an indication indicating the first encoded signal and the second encoded signal; an indication used to decode a first time/frequency resolution of the first encoded signal, and a used to decode the second An indication of the second time/frequency resolution of one of the encoded signals. A computer program for operating on a processor for performing the method of claim 8 or 17.