WO2006011367A1

WO2006011367A1 - Audio signal encoder and decoder

Info

Publication number: WO2006011367A1
Application number: PCT/JP2005/012941
Authority: WO
Inventors: Kazuhiro Iida; Mineo Tsushima; Yoshiaki Takagi; Naoya Tanaka
Original assignee: Matsushita Electric Industrial Co., Ltd.
Priority date: 2004-07-30
Filing date: 2005-07-13
Publication date: 2006-02-02

Abstract

Multichannel signals are easily separated and extracted from a code sequence in which a coded audio signals are mixed without giving the listener a feeling of auditory oddness. When the listener listens to the sound of multichannel signals, the characteristic, a surrounded feeling, is important. Therefore, encoding matching the discrimination characteristic is performed without impairing the characteristic. The surrounded feeling is represented by the ratio Ef/Eb of the sum Ef of the energies of the signals of the channels (L, C, R) in front of the ears of the listener to the sum Eb of the energies of the signals of the channels (BL, BR) behind the ears. In an audio signal encoder of the invention, the energy ratio Ef/Eb of the front and rear energies representing the surrounded feeling is multiplexed in a code sequence in which coded multichannel signals are mixed. As a result, multichannel audio signals giving a feeling of presence can be decoded.

Description

Specification

Audio signal encoding apparatus and decoding apparatus

Technical field

The present invention relates to an audio signal encoding apparatus and decoding apparatus.

Background art

[0002] Conventional audio signal encoding methods and decoding methods include ISO / IEC international standard methods, commonly known MPEG methods, and the like as well-known methods. Currently, ISO / IEC 13818-7, commonly known as MPEG2 AAC (Advanced Audio Coding), is a coding scheme that has a wide range of applications and high sound quality even at low bit rates. Multiple standards for this system are currently being standardized. One of them is a technology that uses information called spatial acoustic information or auditory acoustic information (Binaural Cue). As an example of such a technology, there is a parametric stereo system defined in MPEG-4 Audio (ISO / IEC 14 496- 3), an ISO international standard (for example, see Non-Patent Document 1). ). Another method has also been proposed (see, for example, Patent Document 1 and Patent Document 2).

Non-Patent Document 1: ISO / IEC 14496-3: 2001 AMD2 "Parametric Coding for High Quality Audio

Patent Document 1: US Published Patent US2003 / 0035553 "Backwards- compatible Perceptual Coding of Spatial Cues

Patent Document 2: US Published Patent US2003 / 0219130 "Coherence-based Audio Coding and Synthesis"

Disclosure of the invention

Problems to be solved by the invention

[0003] However, in the conventional audio signal encoding method and decoding method, for example, in the AAC described in the background art, when a multi-channel signal is encoded, the correlation between channels is sufficient. However, it was difficult to achieve a low bit rate due to lack of energy. Even when code sign is implemented using correlation between channels, There was a problem that the effects such as the improvement of the coding efficiency obtained by using the perceptual characteristics related to the feeling of envelopment of the sound source could be fully utilized for quantization and coding.

[0004] Further, in the conventional method, when decoding a multi-channel signal that has been encoded, when playing back with two speakers, headphones, etc., all the channels are once decoded, and thereafter Using a method such as downmixing, it was necessary to generate an audio signal to be reproduced by two speakers and headphones. This required a large amount of calculation and a buffer for calculation when played back with two speakers and headphones, and this caused the power consumption and cost of calculation means such as DSP to be implemented. Means for solving the problem

[0005] In order to solve the above problems, an audio signal encoding device according to the present invention is an audio signal encoding device that encodes a multi-channel audio signal, and is a listener of the multi-channel audio signal. Energy sum calculation means for calculating the sum of the energy of the channel signals existing on the front side and the sum of the energy of the channel signals existing on the back side of the listener, the energy sum on the front side, and the energy on the back side It is characterized by comprising energy ratio calculating means for calculating the ratio to the sum, and energy ratio sign means for signing the calculated ratio of the energy sum.

[0006] The energy ratio sign means may quantize the sign of the energy sum according to a discrimination characteristic related to a sense of wrapping.

[0007] Further, the discrimination feature indicates that the discrimination performance is the highest when the difference between the energy sum on the front side and the energy sum on the back side is within a predetermined range. The ratio sign means that the difference between the energy sums on the front side and the back side is the smallest! / ヽ is quantized so that the quantization accuracy is highest, and the number of bits decreases as the energy sum difference increases. Quantization may be performed.

[0008] The audio signal decoding apparatus of the present invention decodes a code sequence representing a multi-channel audio signal to output a multi-channel audio signal having an auditory wrapping feeling. Audio signal decoding means for generating a multi-channel audio signal by decoding a code string, the signal decoding device By decoding the code string, the ratio between the energy sum of the channel signal existing on the front side of the listener and the energy sum of the channel signal existing on the back side of the listener is decoded. Energy ratio decoding means, and energy distribution means for distributing energy to the front side channel and the back side channel according to the ratio of the decoded energy sum.

The invention's effect

[0009] As described above, in the audio signal encoding method and the decoding method of the present invention, when a plurality of mixed signal sequences are separated into a plurality of signal sequences, a human sound source package is included. It is possible to achieve signal separation to the extent that there is no sense of incongruity in the sense of hearing by generating very small auxiliary information by using the perceptual characteristics related to rare feeling.

[0010] Further, if it is configured to be a signal-mixed multi-channel signal down-mix signal, a down-mix signal that does not read and process the auxiliary information at the time of decoding. If only the part is decoded, even a speaker or headphone having a 2-channel signal playback system can be played back with a low computational complexity and high sound quality.

Brief Description of Drawings

FIG. 1 is a diagram showing a processing flow of an encoding device and a decoding device according to the present embodiment.

FIG. 2 is a diagram showing a relationship between a listener and a sound source indicated by channel information.

[FIG. 3] FIG. 3 is a diagram showing a coding method that changes the quantization accuracy of the front and rear energy ratios based on human auditory characteristics.

FIG. 4 is a block diagram showing a configuration of an encoding device and a decoding device each including the energy ratio encoding unit and the energy distribution unit shown in FIG.

Explanation of symbols

[0012] 100 encoding apparatus

101 Forward channel energy calculator

102 Rear channel energy calculator

103 Front-rear energy ratio calculator

104 Energy quantization code 105 Energy decoding inverse quantization section

106 Front-rear energy calculator

107 Forward channel energy distributor

108 Rear channel energy distributor

109 coded sequence

110 Decryption device

400 Energy ratio sign

401 Downmix

402 Frequency converter

403 Quantizer

404 encoder

405 Multiplexer

406 Demultiplexer

407 Decryption unit

408 Inverse quantization unit

409 Inverse frequency converter

410 Energy distribution department

411 Signal separator

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0014] (Embodiment)

FIG. 1 is a diagram schematically showing a processing flow of audio signal encoding apparatus 100 and decoding apparatus 110 of the present embodiment. In this embodiment, it does not matter what encoding method is used to encode and decode the audio signal of each channel! It will be described later as an example.

[0015] This encoding device 100 is an encoding device that encodes the energy ratio between the front and rear of the multi-channel signal in order to restore the wrapping feeling due to the multi-channel signal in the decoding device 110. In addition, the energy ratio sign 400 is provided. Energy ratio The signal key unit 400 also includes a front channel energy calculation unit 101, a rear channel energy calculation unit 102, a front rear energy ratio calculation unit 103, and an energy quantization code unit 104. The code key sequence 109 is a signal encoded by the code key device 100.

[0016] Also, the decoding device 110 calculates the energy ratio between the front and rear of the multi-channel signal from the code string 109 output by the coding device 100, and outputs each decoded channel. This is a decoding device that distributes energy so that the original signal is wrapped by the signal, and includes an energy distribution unit 410. The energy distribution unit 410 includes an energy decoding / inverse quantization unit 105, a front / rear energy calculation unit 106, a front channel energy distribution unit 107, and a rear channel energy distribution unit 108.

First, the encoding device 100 will be briefly described. Figure 2 shows the relationship between the listener and sound source indicated by the channel information. When assuming a multi-channel audio signal listening system that handles, for example, a 5-channel signal as shown in Fig. 2, the listener is seated facing forward and in front of the listener in front of the listener. The front L channel (L), front R channel (R), center channel (C), and back L channel (BL) and back R channel (BR) located behind the ear are roughly divided into the following processes. To do. The channel ahead of the ear is called the front channel, and the channel behind the ear is called the back channel. For simplicity, the power to explain a system with 5 channels! / A system with other number of channels is possible in the sense that the processing is roughly divided into the front and rear channels from the listener's ear. In general, the LF E (Low Frequency Effect) channel, which handles low frequency signals, is not perceived in the human auditory characteristics even in front of (or behind) the listener. Do not distinguish from channels.

Returning to FIG. 1 again, the front channel signals (L, R, C) are input to the front channel energy calculation unit 101. The front channel energy calculation unit 101 calculates the energy of each of the front channel signals (L, R, C) and calculates the total energy of the front channel signal. The rear channel signals (BL, BR) are input to the rear channel energy calculation unit 102. The rear channel energy calculation unit 102 calculates the energy of each of the rear channel signals (BL, BR) and outputs the rear channel signal. Calculate the total energy. Since the processes of the front channel energy calculation unit 101 and the rear channel energy calculation unit 102 are a part of the audio encoding method, the processes are generally performed at specified time intervals. In other words, energy is calculated at each time interval.

Next, the front channel energy and the rear channel energy are input to the front-rear energy ratio calculation unit 103. The front-rear energy ratio calculation unit 103 calculates the energy ratio between the front channel energy and the rear channel energy. At the same time, the output of the front-rear energy ratio calculation unit 103 includes the absolute amount of energy obtained only by the energy ratio (the front energy itself or the rear energy itself, or the sum of the forward energy and the rear energy. ) Is also output. The energy quantization code unit 104 quantizes the code sequence 109 using the energy ratio and the absolute amount of energy, which are the outputs of the front-rear energy ratio calculation unit 103, as inputs. Here, the quantization method and the sign key method are not particularly specified.

FIG. 4 is a block diagram showing the overall configuration of encoding apparatus 100 and decoding apparatus 110 including energy ratio encoding section 400 and energy distribution section 410 shown in FIG. Encoding apparatus 100 includes an energy ratio encoding unit 400, a downmixing unit 401, a frequency converting unit 402, a quantizing unit 403, an encoding unit 404, and a multiplexing unit 405. Decoding apparatus 110 includes demultiplexing section 406, decoding section 407, inverse quantization section 408, inverse frequency conversion section 409, energy distribution section 410, and signal separation section 411.

First, the encoding device 100 will be described. In parallel with the fact that the energy ratio of the front and rear channels and the absolute amount of energy are quantized and signed in the energy ratio sign section 400, the downmix section 401, the frequency conversion section 402, the quantization section 403 Each channel signal is encoded by the encoding unit 404.

[0022] The front channel signal (L, R, C) and the rear channel signal (BL, BR) force are input to the down-mit- ter unit 401. The downmix unit 401 generates a left downmix signal represented by (L + BL) Z2 from the front left channel signal L and the rear left channel signal BL. Further, a right downmix signal represented by (R + BR) / 2 is generated from the front right channel signal R and the rear right channel signal BR. The center channel signal C remains unchanged. Outputs 3 channel signals. Here, there are a number of proposed methods of down-mittas for explaining the downmixing of (L + BL) Z2 and (R + BR) Z2, and either method may be used. In addition, the present invention is not limited to any downmitas method used here. The signal from the downmix unit 401 is input to the frequency conversion unit 402. The frequency converter 402 converts the signal for each channel into a frequency spectrum on the frequency axis, for example, by a predetermined number of samples. The quantization unit 403 quantizes the frequency-converted signal of each channel. The sign key unit 404 encodes the quantized transform coefficient. Multiplexing section 405 multiplexes the encoded transform coefficients of each channel and the energy ratio encoded by energy ratio encoding section 400 and outputs code string 109.

Next, the decoding device 110 will be described. The decoding apparatus 110 reproduces the energy of the signal of the front channel and the energy of the signal of the rear channel, with the code sequence 109 encoded by the encoding apparatus 100 as an input.

First, the energy decoding / dequantizing unit 105 of the energy distributing unit 410 reads the code sequence separated from the code sequence 109 by the demultiplexing unit 406 shown in FIG. Decodes the channel energy ratio and the absolute amount of energy (front energy itself or rear energy itself, or the sum of forward energy and backward energy). The front / rear energy calculation unit 106 receives the energy ratio of the front channel and the rear channel and the absolute amount of energy, and outputs the energy sum of the front channel and the energy sum of the rear channel. The energy sum of the front channel is input to the front channel energy distribution unit 107, and the energy sum of the rear channel is input to the rear channel energy distribution unit 108.

On the other hand, the demultiplexing unit 406 demultiplexes the code string representing the spectrum of the left channel, the right channel, and the center channel from the code string 109. The decoding key unit 407 decodes a code string corresponding to each channel. Inverse quantization section 408 performs inverse quantization on the decoded spectrum of each channel. The inverse frequency converting unit 409 converts the left channel, right channel, and center channel signals represented by the frequency spectrum into signals represented by a function of time. The signal separator 411 is a downmix converted to a function of time. Separate the signal into the original multi-channel signal. That is, the front left channel signal L and the rear left channel signal BL are separated from the left channel signal, and the front right channel signal R and the rear right channel signal BR are separated from the right channel signal. At this time, the front channel energy distribution unit 107 derives the energy of each of the L, R, and C channels according to the energy ratio between the front channels, and the rear channel energy distribution unit 108 calculates the energy ratio between the rear channels. To derive the energy of each channel of BL and BR. By generating each channel signal according to the energy derived in this way, the wrapping feeling of the original signal by the multi-channel signal is restored.

[0026] A specific code method according to the present invention will be described in more detail with reference to FIG. The encoding apparatus 100 also includes a front channel energy calculation unit 101, a rear channel energy calculation unit 102, a front / back energy ratio calculation unit 103, and an energy quantization code unit 104. The code key sequence 109 is a signal encoded by the code key device 100.

In addition, the decoding device 110 includes an energy decoding / inverse quantization unit 105, a front / back energy calculation unit 106, a front channel energy distribution unit 107, and a back channel energy distribution unit 108.

[0028] The encoding device 100 and the decoding device 110 focus on characteristics related to the feeling of wrapping, which is one of the spatial impressions of listeners when listening to multi-channel audio. , The role of reflections from behind the listener in spatial ι mpression ", Masayu i orimoto, Kazuniro Iida, et.al, Applied Acoustics 2001, pp.1 09-124, etc. It is important to maintain the ratio between the sound source level of the sound source and the sound source level of the rear channel, in other words, the ratio between the sound source level of the front channel and the energy level of the rear channel (in this embodiment, , Which is called FBR (Front Back Energy Ratio)) means that there is a possibility that a listener can be provided with a sufficient feeling of wrapping.

[0029] For simplicity, the front left channel (L), front right channel (R), front center channel (C), back left channel (BL), back right channel ( Assuming 5 channels of (BR), The channels are L, R, C, and the rear channels are BL, BR. A multi-channel system may have a larger number of channels or a smaller number of channels. For the listener, the front channel is the front channel, and the rear channel is the rear channel. It can be handled in the same manner as this embodiment.

First, the front channel signals (L, R, C) are input to the front channel energy calculation unit 101, and the front channel energy calculation unit 101 calculates the total energy of the front channel signal. The rear channel signals (BL, BR) are input to the rear channel energy calculation unit 102, and the rear channel energy calculation unit 102 calculates the total energy of the rear channel signals. Since the processes of the front channel energy calculation unit 101 and the rear channel energy calculation unit 102 are a part of the audio encoding method, the processes are generally performed at specified time intervals. In other words, energy is calculated for each time interval. The energy of the front channel is Ef, and the energy of the rear channel is Eb. For example, if the energies of the front channel signals (L, R, C) are LE, RE, and CE, the total energy Ef of the front channel signals is expressed as Ef = LE + RE + CE. Similarly, if each energy of the rear channel signal (BL, BR) is BLE and BRE, the total energy Eb of the rear channel signal is expressed as Eb = BLE + BRE.

Next, the front channel energy Ef and the rear channel energy Eb are input to the front / rear energy ratio calculation unit 103. The front-rear energy ratio calculation unit 103 calculates the energy ratio FBR (see equation (1)) between the front channel energy and the rear channel energy.

[0032] FBR = 10 1og (Ef / Eb) (1)

At the same time, the output of the front-rear energy ratio calculation unit 103 includes the absolute amount of energy obtained by the above-mentioned energy ratio alone (the front energy Ef itself, or the rear energy Eb itself, or the front energy and the rear energy. The added value Ef + Eb) is also output. The energy quantization code unit 104 receives the energy ratio (FBR), which is the output of the front-rear energy ratio calculation unit 103, and the absolute amount of energy, and inputs the quantity. The code string 109 is generated by the child. As mentioned above, since the energy ratio (FBR) is a value that is strongly related to the listener's audibility characteristics related to the “wrapping feeling”, during quantization, in the range that is sensitively perceived by human auditory characteristics. If the energy ratio (FBR) is finely quantized, that is, quantized with a small quantization step, and coarsely quantized within a less sensitive range, the code efficiency can be increased. FIG. 3 is a diagram showing a coding method that changes the quantization accuracy of the front and rear energy ratios based on human auditory characteristics. In the figure, the horizontal axis represents the front-to-back energy ratio (Ef / Eb), and the internal vertical line represents the density of the quantization accuracy. As shown in the figure, from the viewpoint of human auditory characteristics, the difference between the energy of sound heard from the front and the energy of sound heard from the rear is most perceived when the energy values of the front and rear are the same. Cheap. That is, when Ef I Eb = 1, the difference between the energy values before and after is most easily perceived. On the other hand, if the energy values of the front and back are different from each other originally, even if the original energy is slightly deviated, it is not perceived by the human ear. Therefore, when the front / rear energy ratio (Ef / Eb) is close to “1”, fine quantization is performed for the energy ratio, and coarser quantization is performed as the distance from “1” is increased.

[0033] Further, it is desirable to use a quantization and sign method that aims to maintain the value of the energy ratio (FBR) more than holding the value of the forward energy Ef itself or holding the value of the backward energy Eb itself. Here are the quantization and coding methods! No special provisions ヽ

[0035] First, the code sequence 109 is read by the energy decoding / inverse quantization unit 105, and the energy ratio (FBR) of the front channel and the back channel and the absolute amount of energy ( The forward energy itself Ef, or the backward energy itself Eb, or the sum of the forward energy and the backward energy Ef + Eb) is decoded. The front-rear energy calculation unit 106 receives the energy ratio (FBR) of the front channel and the rear channel and the absolute amount of energy, and calculates the energy sum Ef of the front channel and the rear channel. Output the energy sum Eb. The energy sum Ef of the front channel is input to the front channel energy distribution unit 107, and the energy sum Eb of the rear channel is input to the rear channel energy distribution unit 108. The front channel energy distribution unit 107 derives the energy of each of the channels L, R, and C according to the energy ratio between the front channels, and the rear channel energy distribution unit 108 determines the BL according to the energy ratio between the rear channels. The energy of each channel is derived. The method for decoding the energy ratio between the front channels and the energy ratio between the rear channels is not particularly defined in the present application. In general, decryption is performed based on separately available information.

[0036] If the audio signal encoding device 100 and the decoding device 110 configured in this way are used, it is easy to maintain the characteristic of "envelopment" when listening to the multi-channel of the listener, and there are few It is possible to provide a comfortable sound field even when multi-channel playback is performed by forming a code sequence with information.

Note that the encoding method and decoding method for each channel signal described in the above embodiment are merely examples, and are used in the audio signal encoding device and the decoding device of the present invention. The encoding method and decoding method for each channel signal are not limited to this. For example, in the above example, a 5-channel multi-channel signal was downmixed into 3 channels, the left channel, the right channel, and the center channel, but the left channel, the right channel, and the center channel were encoded. Further, the signal may be downmixed to a monaural signal to be encoded and decoded. Industrial applicability

[0038] The audio signal decoding method and encoding method of the present invention can be applied to all applications to which the audio encoding method and decoding method have been applied.

[0039] Code stream, which is a bit stream encoded with audio code, is currently used for transmission of broadcast contents, applications that are recorded and played back on storage media such as DVDs and SD cards, and communication devices represented by mobile phones. This is used when transmitting AV content to the Internet. In addition, when transmitting audio signals as electronic data exchanged over the Internet. It is useful.

Claims

The scope of the claims

[1] An audio signal encoding device for encoding a multi-channel audio signal,

Energy sum calculating means for calculating the energy sum of the channel signal existing on the front side of the listener and the energy sum of the channel signal existing on the back side of the listener among the multi-channel audio signals;

Energy ratio calculating means for calculating a ratio between the energy sum on the front side and the energy sum on the back side;

An audio signal encoding unit that includes an energy ratio encoding unit that encodes the calculated ratio of the energy sums.

[2] The energy ratio sign means quantizes and encodes the ratio of the energy sums according to a discrimination characteristic related to a sense of wrapping in the auditory sense.

2. The audio signal encoding apparatus according to claim 1, wherein

[3] The discrimination characteristic indicates that the discrimination performance is highest when the difference between the energy sum on the front side and the energy sum on the back side is within a predetermined range.

The energy ratio sign key means quantizes so that the quantization accuracy is highest when the difference in energy sum between the front side and the back side is the smallest, and the number of bits decreases as the difference in energy sum increases. Quantize

3. The audio signal encoding apparatus according to claim 2, wherein

[4] An audio signal decoding apparatus that outputs a multi-channel audio signal having an auditory feeling by decoding a code sequence representing a multi-channel audio signal,

Audio signal decoding means for generating a multi-channel audio signal by decoding the code string;

By decoding the code string, an energy ratio for decoding the ratio between the energy sum of the channel signal existing on the front side of the listener and the energy sum of the channel signal existing on the back side of the listener. Decryption means,

According to the ratio of the decoded energy sum, the front side channel and the bar An energy distribution means for distributing energy to the sound channel;

An audio signal decoding apparatus comprising:

[5] An audio signal encoding method for encoding a multi-channel audio signal,

Among the multi-channel audio signals, calculate the energy sum of the channel signal existing on the front side of the listener and the energy sum of the channel signal existing on the back side of the listener.

Calculate the ratio of the energy sum on the front side and the energy sum on the back side, and sign the ratio of the calculated energy sum

An audio signal encoding method characterized by the above.

[6] An audio signal decoding method for outputting a multi-channel audio signal having an auditory sensation by decoding a code sequence representing a multi-channel audio signal,

By decoding the code string, a multi-channel audio signal is generated, and by decoding the code string, the energy sum of the channel signals existing on the front side of the listener and the back side of the listener exist. Decode the ratio of the signal of the channel to be

Distributes energy to the front channel and the back channel according to the ratio of the decoded energy sum

An audio signal decoding method characterized by the above.

[7] A program for an audio signal encoding device, which is a multi-channel audio signal, a sum of energy of signals of channels existing on the front side of the listener, and a channel existing on the back side of the listener. A step of calculating an energy sum of the signals, a step of calculating a ratio of the energy sum of the front side and the energy sum of the back side, and a step of signing the ratio of the calculated energy sum A program that causes a computer to execute each step it contains.

[8] A program for an audio signal decoding apparatus, which generates a multi-channel audio signal by decoding a code string, and decodes the code string By decoding, the step of decoding the ratio of the energy sum of the signal of the channel existing on the front side of the listener and the energy sum of the signal of the channel existing on the back side of the listener, and the decoded energy A program that causes a computer to execute each step including a step of distributing energy to the front channel and the back channel according to a ratio of the sum.

[9] Among the multi-channel audio signals, a step of calculating an energy sum of channel signals existing on the front side of the listener and an energy sum of channel signals existing on the back side of the listener; A program for causing a computer to execute each step including a step of calculating a ratio between the energy sum and the energy sum on the back side and a step of signing the calculated ratio of the energy sums is recorded. Computer-readable recording medium.

[10] A step of generating a multi-channel audio signal by decoding the code sequence, and an energy sum of the channel signals existing on the front side of the listener by decoding the code sequence And a step of decoding the ratio of the energy of the signal of the channel existing on the back side of the listener, and the energy of the front side channel and the back side channel according to the ratio of the decoded energy sum A computer-readable recording medium having recorded thereon a program for causing a computer to execute each step including the step of distributing the program.