DE112008003153B4 - Frequency band determination method for shaping quantization noise - Google Patents

Abstract

A frequency band determination method for shaping quantization noise, comprising: checking whether audio signals obtained from low-pass filtering are transient; Determining a predetermined frequency band to apply as a frequency band to be used for shaping quantization noise when the audio signals are not transient; and determining an extended frequency band extended beyond the predetermined frequency band to apply as a frequency band to be applied when the audio signals are transient, wherein the predetermined frequency band to be applied is a specific TNS frequency band a general TNS algorithm is applied, and wherein the extended frequency band is extended down to a frequency band lower than a certain TNS frequency band.

Description

Technical area

The present invention relates to a frequency band determination method for shaping quantization noise. In particular, the method shapes the noise using a long block and reduces pre-echo and musical noise by calculating whether an applied frequency band is a general frequency band and an extended frequency band based on the fact whether audio signals are transient signals.

This work was supported by the IT R & D Program for MIC / IITA [2007-S-005-01, "Development of Rich Media Broadcasting Technologies through Advanced Audio and Video Codec Technologies"].

Technical background

In a highly efficient advanced audio coding (HE-AAC) technology, a temporal noise shaping (TNS) algorithm, which is one of quantization noise shaping algorithms, is used to effectively represent transient signals. Therefore, the TNS algorithm reduces a pre-echo.

Despite the use of the TNS algorithm, however, at low bit rates, the pre-echo and the musical noise often occur.

In order to capture sound quality with perceptibly transparent coding in HE-ACC audio coding technology, quantization noise should not exceed a masking threshold. However, in a perceptual coding technology using a frequency signal analysis method, the quantization noise is coded and then widely spread in the time domain. Therefore, at low bit rates, it is difficult to satisfy the condition that the quantization noise does not exceed the masking threshold in the time domain.

For example, in AAC audio coding technology, which generally uses 1024 Modified Discrete Cosine Transform (MDCT) coefficients, quantization noise is distributed over a 48 ms sampling rate over 40 ms. This distribution can cause audible artifacts when the signals are transient signals. At this time, the quantization noise may be perceptibly detected before the transient signals are generated. This quantization noise is called the pre-echo phenomenon.

In the TNS algorithm which has been developed to deal with the pre-echo phenomenon effectively, the shape of quantization noise that is widely spread in the time domain is adjusted to have a masking effect.

The TNS algorithm uses Linear Predictive Coding (LPC) based on a duality between the time domain and a frequency domain. Table 1 illustrates an optimal encoding method for audio signals and transient signals with respect to duality.

Table 1

That is, in the frequency domain, the optional coding method for the tone signals having a certain frequency is a direct coding method using frequency coefficient coding. In the time domain, the optimal coding method for the tone signals having a particular frequency is a predictive coding method using LPC coding.

On the other hand, considering the duality, in the frequency domain, the optimal coding method for the transient signals is the predictive coding method using the frequency coefficient prediction coding. In the time domain, the optional encoding method for the transient signals is the direct encoding method that uses time sampling coding.

The TNS algorithm is applied using the frequency domain predictive coding technique.

Table 2 shows the frequency band that TNS applies. Table 2 Types of blocks long block short block Frequency band applying TNS 1275 Hz ~ 2750 Hz ~

Here, the TNS frequency range (band) is classified according to the block length into a long block and a short block. The frequency band using the TNS algorithm is more than 1275 Hz for the long block and more than 2750 Hz for the short block.

That is, in the long block, the TNS algorithm is applied to a frequency range from a frequency band of 1275 Hz to a cut-off frequency at which spectral band replication (SBR) begins. On the other hand, in the short block, the TNS algorithm is applied to a frequency range from a frequency band of 2750 Hz up to the cutoff frequency at which the SBR starts. In a band below the above frequency bands, the pre-echo occurs frequently.

To further reduce the pre-echo, a block switch is executed. Block switching refers to the process of replacing a long window with a frame length with a 1/8 frame length short window. The block switching between the long block and the short block serves to perceptibly improve the pre-echo by applying the quantization noise effect only in the short block.

However, if the signals are stable and the bit rate is low, the short window may produce an adverse effect. Since the bit in the low bit rate is insufficient, frequency components lost in each short block are shown as spectral holes. The spectral holes are intermittently connected on a time axis in a corresponding frame to cause the musical noise. That is, at the low bit rate with insufficient bit, when the long block is used instead of the short block, the pre-echo occurs. In addition, at low bit rate with insufficient bit, when the short block is excessively used, the musical noise occurs.

The publication US 2007/0016405 A1 describes encodings with improved time resolution for selected segments by means of adaptive block transformation of a group of signal values of a subband decomposition.

Jürgen Herre, "Temporal Noise Shaping, Quantization, and Coding Methods in Perceptual Audio Coding: A Tutorial Introduction", in: AES 17th International Conference: High Quality Audio Coding; Publication Date: August 1, 1999; describes a method of predictive coding applied to spectral data versus frequency, wherein the temporal shape of the quantization error signal at the output of the decoder appears to be matched to the temporal shape of the input signal.

Disclosure of the invention

Technical problem

An embodiment of the present invention is directed to a frequency band determination method for shaping quantization noise.

In the conventional Temporal Noise Shaping (TNS) technology, when the bit rate is low, pre-echo and musical noise in transient blocks can not be prevented. An object of the present invention is to solve the above problems.

Thus, this invention provides a method of determining a frequency range for forming quantization noise using a long block in accordance with an applied frequency band classified as a general frequency band or an extension frequency band, and depending on whether audio signals are transient signals to effectively reduce the pre-echo and the musical noise, and a method of shaping transient noise using this method.

The objects of the present invention are not limited to the above-mentioned objects. Other objects and advantages of the present invention can be understood by the following description, and become apparent with reference to the embodiments of the present invention. Moreover, it will be apparent to those skilled in the art that objects and advantages of the present invention may be realized by the means claimed and combinations thereof.

Technical solution

According to one aspect of the present invention, there is provided a frequency band determining method for shaping quantization noise, comprising: checking whether audio signals obtained from low-pass filtering are transient signals, determining a predetermined frequency band to apply as a frequency band thereto the shaping of the quantization noise is to be applied when the audio signals are non-transient and determining an extended frequency band extended beyond the predetermined frequency band to apply as a frequency band to be applied when the audio signals are transient, wherein the predetermined frequency band which is to be applied is a particular TNS frequency band applied to a general TNS algorithm, and wherein the extended frequency band is extended down to a frequency band lower than a certain TNS frequency band.

Advantageous effects

This invention forms quantization noise of audio signals by using a long block in accordance with a frequency band for the application of an algorithm for shaping temporal noise (TNS algorithm) and classifies the frequency band into a general frequency band and an extension frequency band in accordance with whether the audio signals are transient signals. Thus, a pre-echo and musical noise can be easily reduced.

Therefore, in this invention, the sound quality is better than a method of shaping quantization noise using the conventional TNS algorithm. The method of the present invention using the long block reduces the pre-echo more effectively than the typical method. Further, it is possible to provide almost the same performance that can be obtained in the method using only the short block.

Therefore, in this invention, the short block is not excessively used, whereby the musical noise can be reduced.

Brief description of the drawings

1 Fig. 10 is a block diagram showing an apparatus for processing the shaping of temporal noise (TNS processing) according to an embodiment of the present invention.

2 and 3 show the pre-echo according to the transient index.

4 FIG. 10 is a flowchart describing a method of shaping quantization noise in a low frequency band using a long block according to an embodiment of the present invention. FIG.

Best way to carry out the invention

The advantages, features and aspects of the invention will become apparent from the following description of the embodiments, which refers to the accompanying drawings and is given below. Therefore, those skilled in the art can readily understand the technological concept and scope of the invention. Further, assuming that a detailed description of a related art could obscure a point of the present invention, the detailed description will not be given here. The preferred embodiments of the present invention will be described below in detail with reference to the accompanying drawings.

1 Fig. 10 is a block diagram showing an apparatus for processing the shaping of a temporal noise (TNS processing) according to an embodiment of the present invention.

In 1 contains a TNS processing device 100 a TNS determination device 110 and a TNS encoder 120 ,

The TNS processing device 100 reshapes the quantization noise in a time domain in a filter bank window to make the noise imperceptible. Hereinafter, the TNS processing device 100 in a general HE-ACC encoding device.

The TNS determining device 110 determines whether a TNS process is applied or not.

wobei k und n eine MDCT-Koeffizienteneinheit bzw. eine Skalenfaktoreinheit darstellen.More specifically, the TNS determiner multiplies 110 a weight as in Eq. 1 to calculate a linear predictive coding (LPC) of a pre-computed modified discrete cosine transform (MDCT) spectrum. X _w (k) = X (k) · WFAC (k) (Eq. 1) in which

where k and n represent an MDCT coefficient unit and a scale factor unit, respectively.

This means that Eq. 1 normalized to the energy of a corresponding scale band. The MDCT spectrum range is applied to a given range. Therefore, the TNS determining means determines 110 a frequency range (band) for the application of the LPC.

wobei k und n eine MDCT-Koeffizienteneinheit bzw. eine Skalenfaktoreinheit darstellen.The TNS determination unit 110 applies a smoothing filter to the normalized spectrum. This is for the LPC analysis. The smoothing filtering means down-filtering in a frequency band range from an LPC interrupt frequency to an LPC operating frequency by the one shown in Eq. 2 shown process.

where k and n represent an MDCT coefficient unit and a scale factor unit, respectively.

wobei k und n eine MDCT-Koeffizienteneinheit bzw. eine Skalenfaktoreinheit repräsentieren.On the other hand, the TNS determining means performs 110 up-filtering in a frequency band range from the LPC operating frequency to a cut-off frequency. Eq. 3 shows the up-filtering process.

where k and n represent an MDCT coefficient unit and a scale factor unit, respectively.

The TNS determining device 110 calculates an autocorrelation function and the LPC using a Levinson-Durbin algorithm. The TNS determining device 110 detects a Partial Autocorrelation Coefficient (PARCOR) and calculates a prediction gain based on the result of the calculation using the Levinson-Durbin algorithm.

If the calculated prediction gain exceeds a threshold, the TNS determiner determines 110 that the LPC should be performed on the spectrum and that a TNS algorithm should be applied to the current window.

The TNS encoder 120 performs a quantization simulation in an order of high to low PARCOR coefficients to determine the TNS order and a first coefficient not less than the threshold, e.g. B. 0.1, is. This is to use only effective TNS PARCOR coefficients.

The TNS encoder 120 passes through an LPC filter of the particular order and coefficient and applies the TNS algorithm to the MDCT spectral coefficients to perform the encoding. The ACC coding is performed using the applied MDCT spectral coefficient.

This invention extends the TNS algorithm down to low frequencies, such as 100 Hz, and applies it to it. Since the TNS algorithm is applied here extended, the pre-echo is reduced. The sound components of the frequency applying the TNS algorithm, d. H. the low frequencies, however, can be distorted.

Therefore, this invention simultaneously uses the general TNS algorithm and the extension TNS algorithm. This means that this invention determines whether it is the general TNS Apply algorithm or the extension TNS algorithm, whereupon it executes the TNS algorithm based on the determination result. A reference for the determination is the extended low frequency extension range.

The following describes conditions for determining whether the TNS algorithm or the extension TNS algorithm is used.

First, the TNS determiner determines 110 whether the general TNS algorithm can be applied or not. If the block switch result is transient only in the low frequency band, the TNS determiner applies 110 the extension TNS algorithm.

Second, the TNS determiner applies 110 the extension TNS algorithm when the prediction gain in the frequency band applying the TNS algorithm extended up to 100 Hz exceeds the threshold and the transient signals with increased energy are between the fourth and seventh frames of eight frames. On the other hand, the TNS determining means applies 110 the general TNS algorithm when the transient signals with reduced energy are between the zeroth and the third frame of eight frames.

The transient index of the transient signals 0 to 7 indicates the transient index determined by the block switching between the short block and the long block. Each block indicates the individual points at which the corresponding frame is divided into eight frames. This transient index is used for the effective coding for the HE-ACC and is referred to when eight short blocks are tied together into four groups to apply the short blocks. The reference value for the TNS algorithm is the transient index of the signals sent and filtered by the low frequency.

Under the second condition described above, the effect of the pre-echo is considered and the extent of the effect affects the extension TNS application. Here, the time for the pre-echo is equal to the corresponding window length.

That is, when the transient high-energy portion is in a front portion of the corresponding frame, the pre-echo occurs in a narrower range. On the other hand, when the transient high energy portion is in an end portion of the corresponding frame, the pre-echo occurs in a wider range.

The 2 and 3 show the pre-echo according to the transient index.

In 2 gives a first transient index 101 a transient signal in an end portion of the frame. Because here is the first pre-echo 102 located in the end portion of the frame, the first pre-echo occurs 102 in a larger area.

In 3 gives a second transient index 103 a transient signal in a front portion of the frame. If here is the second transient index 103 located in the end portion of the frame has the second pre-echo 104 a greater impact than the first pre-echo 102 , this in 2 is shown. Therefore, the TNS determining means determines 110 that the extension DNS algorithm is applied.

4 Fig. 10 is a flowchart describing a method of shaping the quantization noise in a low frequency band using a long block according to an embodiment of the present invention.

A TNS determination device 110 calculates a prediction gain of audio signals using a long block in step S302. That is, the TNS determiner 110 calculates the autocorrelation function and the LPC using a Levinson-Durbin algorithm, and obtains a PARCOR from the computation result and calculates a prediction gain.

The TNS determining device 110 determines in step S304 whether the calculated prediction gain exceeds the threshold.

wobei H(z) die Tiefpassfilterfunktion angibt. Auf H(z) können verschiedene Tiefpassfilterfunktionen angewendet werden. In der Tiefpassfilterungsfähigkeit der verschiedenen Tiefpassfilterfunktionen besteht kein großer Unterschied. Die TNS-Bestimmungseinrichtung 110 verwendet das Tiefpassfilter, um das Signal in einem niedrigen Frequenzband unter 1 kHz zu erfassen. The TNS determining device 110 Separately, perform low pass filtering using the low pass filter to determine frequency components in an extension band. This is to use only the long block. An example of the low pass filter function is given in Eq. 4.

where H (z) indicates the low pass filter function. On H (z) various low pass filter functions can be applied. There is not much difference in the low pass filtering capability of the various low pass filter functions. The TNS determining device 110 uses the low-pass filter to capture the signal in a low frequency band below 1 kHz.

Based on the comparison result in step S304, the TNS determining means checks 110 if the prediction gain exceeds the threshold, in step S306, whether the low-pass filtered and filtered signals are transient. That is, the TNS determiner 110 determines, in accordance with the result of the check in step S306, the frequency band used for shaping the quantization noise. On the other hand, if the prediction gain does not exceed the threshold, the TNS determiner calculates 110 in step S314, the prediction gain of the expansion band and checks whether the prediction gain of the expansion band exceeds the threshold. Here, it is checked whether the signals sent and filtered by the low frequency pass are transient by using the block switching algorithm in the AAC device.

Based on the result of the check in step S306, the TNS determining means performs 110 then, when the signals sent and filtered by the low frequency are transient, no re-adjustment of a masking threshold 308 is made. On the other hand, if the signals sent and filtered by the low frequency are not transient, the TNS determiner determines 110 in that the general TNS algorithm is applied, in step 312.

If the masking threshold is not reset, in the general TNS algorithm, the threshold in the frequency band that does not use the TNS algorithm is lowered to effectively use the bit. On the other hand, in this invention using the extension TNS algorithm, since all frequency bands apply the TNS algorithm here, re-adjustment of the masking threshold is not required.

The TNS determining device 110 does not reset the masking threshold and in step S310 expands the frequency band using the TNS algorithm down to the frequency of about 100 Hz. The TNS Encoder 120 recalculates the coefficient based on the TNS algorithm which is extended to the low frequency band, whereupon it performs the TNS coding.

Based on the check result in step S314, the TNS determining means analyzes 110 when the prediction gain of the extension band exceeds the threshold, the type and index of the transient signal sent and filtered by the low frequency, and checks in step S316 if the effect of the pre-echo exceeds a reference value. That is, the TNS determiner 110 determines whether or not the quantization noise is formed based on the analysis result in step 316. For example, the TNS determining means determines 110 in that the effect of the pre-echo exceeds the reference value, and applies the TNS algorithm if the type and index of the transient with increased energy are in the end portion of the corresponding frame, or if the type and index of the low-energy transient in the front section of the corresponding frame.

On the other hand, if in step S314 the prediction gain of the extension band based on the test result does not exceed the threshold, the TNS determiner applies 110 the TNS is not on in step S318.

The TNS determining device 110 does not reset the masking threshold in step S308 if, based on the result of the check in step S316, the effect of the pre-echo is large. On the other hand, if the effect of the pre-echo is small, the TNS determiner applies 110 do not start the TNS algorithm in step S318.

As described above, the technology of the present invention can be realized as a program. A computer programmer of the related art can easily infer a code and a code segment that make up the program. In addition, the implemented program is in a computer-readable recording medium, i. H. stored in an information storage medium and is read and executed by the computer, thereby realizing the method of the present invention. The recording medium includes all types of recording media that can be read by the computer.

The present application contains the subject of the invention Korean Patent Application Nos. 2007-0119413 and 2008-0048837 , filed with the Korean Patent Office on November 21, 2007 and May 26, 2008, respectively, the entire contents of which are incorporated herein by reference.

Although the present invention has been described with respect to certain preferred embodiments, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the scope of the invention as defined in the following claims.

Claims (3)

A frequency band determination method for shaping quantization noise, comprising: Checking whether audio signals obtained from low-pass filtering are transient; Determining a predetermined frequency band to apply as a frequency band to be used for shaping quantization noise when the audio signals are not transient; and Determining an extended frequency band extended beyond the predetermined frequency band to apply it as a frequency band to be applied when the audio signals are transient, wherein the predetermined frequency band to be applied is a particular TNS frequency band applied to a general TNS algorithm, and wherein the extended frequency band is extended down to a frequency band lower than a certain TNS frequency band. The method of claim 1, further comprising: Comparing a prediction gain of the audio signals using a long block with a threshold value; Checking if a prediction gain of the expansion band exceeds the threshold if the prediction gain of the audio signals does not exceed the threshold; Examining energy and arrangement of transient signals of the audio signals; and Determining the extended frequency band as a frequency band to be applied when the prediction gain of the frequency band exceeds the threshold and the transient signals of the enhanced energy audio signals are located in an end portion of the corresponding frame. wherein the checking as to whether audio signals obtained from low-pass filtering are transient is performed when the prediction gain exceeds the threshold value. The method of claim 2, further comprising: Determining the extended frequency band as a frequency band to be applied when the prediction gain of the frequency band exceeds the threshold and the transient signals of the reduced energy audio signals are located in an initial portion of the corresponding frame.

Images