EP2633520B1

EP2633520B1 - Parametric encoder for encoding a multi-channel audio signal

Info

Publication number: EP2633520B1
Application number: EP10859153.8A
Authority: EP
Inventors: Christof Faller; Lei Miao; Yue Lang; Jianfeng Xu
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2010-11-03
Filing date: 2010-11-03
Publication date: 2015-09-02
Anticipated expiration: 2030-11-03
Also published as: CN102844808B; EP2633520A1; ES2553398T3; EP2633520A4; WO2012058805A1; CN102844808A

Description

FIELD OF THE INVENTION

The present invention relates to audio coding.

BACKGROUND OF THE INVENTION

Parametric stereo or multi-channel audio coding as described e.g. in C. Faller and F. Baumgarte, "Efficient representation of spatial audio using perceptual parametrization," in Proc. IEEE Workshop on Appl. of Sig. Proc. to Audio and Acoust., Oct. 2001, pp. 199-202, uses spatial cues to synthesize down-mix - usually mono or stereo - audio signals to signals with more channels. Usually, the down-mix audio signals result from a superposition of a plurality of audio channel signals of a multi-channel audio signal, e.g. of a stereo audio signal. These less channels are waveform coded and side information, i.e. the spatial cues, relating to the original signal channel relations is added as encoding parameters to the coded audio channels. The decoder uses this side information to re-generate the original number of audio channels based on the decoded waveform coded audio channels.
A basic parametric stereo coder may use inter-channel level differences (ILD) as a cue needed for generating the stereo signal from the mono down-mix audio signal. More sophisticated coders may also use the inter-channel coherence (ICC), which may represent a degree of similarity between the audio channel signals, i.e. audio channels. Furthermore, when coding binaural stereo signals e.g. for 3D audio or headphone based surround rendering, also an inter-channel phase difference (IPD) may play a role to reproduce phase/delay differences between the channels.
The synthesis of ICC cues may be relevant for most audio and music contents to re-generate ambience, stereo reverb, source width, and other perceptions related to spatial impression as described in J. Blauert, Spatial Hearing: The Psychophysics of Human Sound Localization, The MIT Press, Cambridge, Massachusetts, USA, 1997. Coherence synthesis may be implemented by using de-correlators in frequency domain as described in E. Schuijers, W. Oomen, B. den Brinker, and J. Breebaart, "Advances in parametric coding for high-quality audio," in Preprint 114th Conv. Aud. Eng. Soc., Mar. 2003. However, the known synthesis approaches for synthesizing multi-channel audio signals may suffer from an increased complexity. Furthermore, the use of ICC parameters, e.g. in addition to other parameters, such as inter-channel level differences (ICLDs) and inter-channel phase differences (ICPDs), may increase a bitrate overhead.
US 2005/180579 A1 discloses a scheme for stereo and multi-channel synthesis of inter-channel correlation (ICC) (normalized cross-correlation) cues for parametric stereo and multi-channel coding. The scheme synthesizes ICC cues such that they approximate those of the original. For that purpose, diffuse audio channels are generated and mixed with the transmitted combined (e.g., sum) signal(s). The diffuse audio channels are preferably generated using relatively long filters with exponentially decaying Gaussian impulse responses. Such impulse responses generate diffuse sound similar to late reverberation. An alternative implementation for reduced computational complexity is proposed, where inter-channel level difference (ICLD), inter-channel time difference (ICTD), and ICC synthesis are all carried out in the domain of a single short-time Fourier transform (STFT), including the filtering for diffuse sound generation.
WO 2003/219130 A1 discloses a combination device that includes: a detection unit that detects active coded bitstreams that are effective coded bitstreams from a plurality of coded bitstreams within a predetermined time period; a first combining unit that combines, from a plurality of downmix sub-streams included in the coded bitstreams, only downmix sub-streams included in the active coded bitstreams so as to generate a combined downmix sub-stream; and a second combining unit that combines, from a plurality of parameter sub-streams included in the coded bitstreams, only parameter sub-streams included in the active coded bitstreams so as to generate a combined parameter sub-stream.
US 2003/219130 A1 discloses an auditory scene synthesized from a mono audio signal by modifying, for each critical band, an auditory scene parameter (e.g., an inter-aural level difference (ILD) and/or an inter-aural time difference (ITD)) for each sub-band within the critical band, where the modification is based on an average estimated coherence for the critical band. The coherence-based modification produces auditory scenes having objects whose widths more accurately match the widths of the objects in the original input auditory scene.

SUMMARY OF THE INVENTION

A goal to be achieved by the present invention is to reduce complexity of a parametric coding scheme. This goal is achieved by the features of the independent claims. Further embodiments are apparent from the description, the drawings and from the dependent claims.
The invention is based on the finding that combining parametric encoding parameters such as ICC parameters may reduce bit rate required for representing the parameters and thus may reduce complexity of the resulting parametric encoding scheme. The combined encoding parameters may be applied e.g. only to a certain frequency region in order to improve an audio quality for e.g. speech whereby the complexity and the memory requirements may further be reduced.
The invention is described in the independent claim 1 and 6. Further embodiments are defined in the dependent claims 2-5
According to a first implementation form, the first and second encoding parameter may be an inter-channel phase difference.
According to a second implementation form, the first and second encoding parameter may be an inter-channel coherence.
According to a third implementation form, the first and second encoding parameter may be an inter-channel intensity difference.
According to a fourth implementation form, the first and second encoding parameter may be an inter-channel level difference.
According to a firth implementation form, the parameter generator is configured to generate the first encoding parameter and the second encoding parameter upon a basis of a multiplication of values of the first transformed audio signal and of the second transformed audio signal.
According to a sixth implementation form, the parameter combiner is configured to determine a weighted average of the first encoding parameter and the second encoding parameter using powers of the a first transformed audio signal and the second transformed signal at the certain frequency as weights to obtain the combined encoding parameter.
According to a seventh implementation form, the parameter combiner is configured to determine a weighted average of the first encoding parameter and the second encoding parameter using a frequency-dependent weight to obtain the combined encoding parameter.
According to an eighth implementation form, the parameter generator is configured to generate a plurality of encoding parameters from the first transformed audio signal and from the second transformed audio signal at a plurality of frequencies, and wherein the parameter combiner is configured to combine the plurality of the encoding parameters to obtain the combined encoding parameter.
According to a ninth implementation form, the parametric encoder further comprises a signal combiner for combining the first transformed audio signal and the second transformed audio signal to obtain a down-mix signal.
According to a tenth implementation form, the parametric encoder further comprises an inverse transformer for inversely transforming a combination of the first transformed audio signal and the second transformed audio signal to obtain a down-mix audio signal.
According to a second aspect the invention relates to a method for parametrically encoding a multi-channel audio signal having a first audio signal and a second audio signal, the method having transforming the first audio signal into frequency domain to obtain a first transformed audio signal, and transforming the second audio signal into frequency domain to obtain a second transformed audio signal, generating a first encoding parameter from the first transformed audio signal and from the second transformed audio signal at a first frequency, and generating a second encoding parameter from the first transformed audio signal and from the second transformed audio signal at a second frequency, and combining the first encoding parameter and the second encoding parameter to obtain a combined encoding parameter.
Further method steps of implementation forms according to the second aspect are directly derivable from the functionality of the parametric encoder according to the first aspect.

BRIEF DESCRIPTION OF THE DRAWINGS

Further embodiments of the invention will be described with reference to the following drawings, in which:

Fig. 1 shows a block diagram of a parametric encoder according to an implementation form;
Fig. 2 shows a block diagram of a parametric decoder according to an implementation form;
Fig. 3 shows a diagram of a method for parametrically encoding according to an implementation form; and
Fig. 4 shows a diagram of a method for parametrically decoding according to an implementation form.

DETAILED DESCRIPTION OF THE EMBODIMENTS

Fig. 1 shows a diagram of a parametric encoder for encoding a multi-channel audio signal having a first audio signal, x1, and a second audio signal, x2, according to an implementation form. The parametric encoder comprises a transformer 101 for transforming the first audio signal into frequency domain to obtain a first transformed audio signal, and for transforming the second audio signal into frequency domain to obtain a second transformed audio signal. The transformer 101 may comprise a first transformer 103 for transforming the first audio signal, and a second transformer 105 for transforming the second audio signal. The transformer 101 and/or the transformers 103, 105 may be Fourier transformers, by way of example. The first and second transformed audio signals are provided to a parameter generator 107 for generating a first encoding parameter from the first transformed signal and from the second transformed audio signal at a first frequency, e.g. at the i-th frequency or in the i-th band. The i-th band or "band i" (see also in Fig. 1) refer to a frequency band i at or in which the parameter generator 107 generates the respective encoding parameter from the first and second transformed signal, and is also referred to as parameter band i. The parameter generator 107 is further configured to generate a second encoding parameter from the first and second transformed audio signal at a second frequency or in a second band. The first and the second encoding parameters are provided to the parameter combiner 109 which combines the first encoding parameter and the second encoding parameter to obtain a combined encoding parameter according to a principle described herein. However, the parameter combiner 109 may separately obtain encoding parameters for different parameter bands.
With reference to Fig. 1 and to ICC parameters forming an embodiment of encoding parameters, the e.g. stereo input audio channels x1 and x2 are converted to a plurality of sub-bands or parameter bands. In all or in a subset of the parameter bands the corresponding ICC parameters may be estimated. One or more ICC parameter combining processes, e.g. one of the processes according to the equations (1)-(4), may be applied to the ICC parameters of all or subsets of parameter bands, to compute the combined ICC parameters. At least one combined ICC parameter may be put into a bit stream 111 or transmitted to an audio decoder which is not depicted in Fig. 1.
The following embodiments are exemplarily described with respect to ICC forming an embodiment of an encoding parameter. It is, however, to be understood, that the encoding parameter may be any encoding parameter or of any encoding parameter type used for parametric encoding, e.g. inter-channel phase difference or inter-channel intensity difference or an inter-channel level difference or the like, and that the encoder may be adapted to produce combined encoding parameters according one, some or all of the aforementioned encoding parameter types and to include combined encoding parameters of different types in the bitstream 111 as side information.
The parametric encoder of Fig, 1 may form a parametric stereo encoder which estimates in parameter bands perceptual spatial cue parameters, such as ICLD, ICPD, and/or ICC. If the parameter band index is i, then the estimated parameters in that band are denoted ICLD(i), ICPD(i), and ICC(i). The left and right signal power in a parameter band are denoted P1(i) and P2(i), respectively.
In this regard, one or more combined ICC parameters may be computed, e.g. as an average $ICC = \frac{1}{N_{I}} \sum_{i \in I} ICC (i)$

where I is the set of indices of parameter bands of which ICC are used to compute the combined ICC parameter and NI is the number of indices in the set I.
Another way of computing a combined ICC parameter is to use a weighted average, i.e. $ICC = \frac{\sum_{i \in I} (P_{1} (i) + P_{2} (i)) ICC (i)}{\sum_{i \in I} (P_{1} (i) + P_{2} (i))}$

wherein P1(i) denotes a signal power of the first audio channel signal in the i-th band, and wherein P2(i) denotes a signal power of the second audio channel signal in the i-th band.
Additionally, different parameter bands (frequencies) can be weighted differently, when computing the combined ICC: $ICC = \frac{\sum_{i \in I} g_{i} (P_{1} (i) + P_{2} (i)) ICC (i)}{\sum_{i \in I} g_{i} (P_{1} (i) + P_{2} (i))}$

where g_i is a weight given to frequency (parameter band) i.
Another example is to use an average considering not power but frequency weighting: Additionally, different parameter bands (frequencies) can be weighted differently, when computing the combined ICC: $ICC = \frac{\sum_{i \in I} g_{i} ICC (i)}{\sum_{i \in I} g_{i}}$
A single full-band ICC performs surprisingly well. In this case, the combined ICC is computed using all parameter bands, i.e. I contains all parameter band indices.
According to some implementation forms, the speech quality may be improved by only using ICC in a limited frequency range. When only parameter bands between 500 Hz and 1.5 kHz are used for generating ICC parameters, less artifacts occur. In this case, the combined ICC is computed using only parameter bands between 500 Hz and 1.5 kHz, i.e. I contains only those indices.
According to some implementation forms, the parametric encoder shown in Fig. 1 may estimate one or more combined ICC parameters by:

(a) Combining the ICC parameters from a plurality of parameter bands to a combined ICC parameter,
(b) Putting combined ICC parameters into a bit stream, and
(c) Outputting the bit stream.

Fig. 2 shows a block diagram of a parametric encoder for decoding a down-mix audio signal according to an implementation form. The down-mix signal may be provided by the parametric encoder as shown e.g. in Fig. 1. The parametric decoder comprises a transformer 201 for transforming the down-mix audio signal, as, to obtain a transformed down-mix audio signal having a certain frequency, e.g. an i-th frequency of a plurality of frequencies, or correspondingly a certain band, e.g. an i-th band of a plurality of bands. The parametric decoder further comprises a provider 203 for providing a frequency-specific encoding parameter associated with the certain frequency. The frequency-specific encoding parameter may be derived from the combined encoding parameter. However, the frequency-specific encoding parameter may correspond to the combined encoding parameter. The parametric decoder further comprises an audio synthesizer 205, e.g. a stereo synthesizer, for synthesizing a first audio signal and a second audio signal at the certain frequency or in the certain band from the transformed down-mix audio signal provided by the transformer 201 using the frequency-specific encoding parameter as provided by the provider 203.
According to some implementation forms, the transformer 201 may be a Fourier transformer, wherein the audio synthesizer may synthesize the first and the second audio signal in frequency domain. Thus, the output signal provided by the synthesizer 205 may correspond to the first and second audio channel signal. However, according to some implementation forms, the parametric encoder may further comprise an inverse transformer 207 for inversely transforming the first and second audio channel signal in time domain in order to obtain a first and second audio channel signal, x1 and x2, in time domain.
The parametric decoder shown in Fig. 1 uses for all parameter bands, or a subset J thereof, the combined ICC parameter or a modified version thereof. One can use the same subset of parameter bands at the decoder as at the encoder,
i.e. J=I, or a different subset.
With respect to Fig. 2, the parametric decoder may receive the down-mix signal s and the stereo parameters, i.e. encoding parameters, amongst which at least one combined ICC parameter may be received. For at least one parameter band an ICC parameter derived from combined ICC parameters is used. To some bands no ICC synthesis may be applied.
According to an implementation form, for the "Combined ICC to Band ICC Conversion" when one combined ICC is used, combined ICC is applied to all parameter bands. Or, if the combined ICC was estimated only for a subset of bands, J, then the decoder may apply the combined ICC to all bands, to the same subset, or to another subset, e.g. a subset of the same subset.
If two combined ICC are used, representing two distinct frequency regions of the audio signal. The decoder may apply the combined ICCs to parameter bands corresponding to the frequency regions from which the combined ICCs were estimated.
Fig. 3 shows a diagram of a method for parametrically encoding a multi-channel audio signal having the first and the second audio signal as mentioned above. The method comprises transforming 301 the first and second audio signal into frequency domain to obtain a first and second transformed audio signal, generating 303 a first encoding parameter from the first and second transformed audio signal at a first frequency, and a second encoding parameter from the first and second transformed audio signal at a second frequency, and combining 305 the first and second encoding parameter to obtain a combined encoding parameter. By way of example, the method depicted in Fig. 3 may be performed by the parametric encoder as shown in Fig. 1.
Fig. 4 shows a block diagram of a method for parametrically decoding a down-mix audio signal upon a basis of a combined encoding parameter. The down-mix audio signal may represent a combination, e.g. a superposition, of a first and second audio signal. The combined encoding parameter may have features as described above.
The method comprises transforming 401 the down-mix audio signal to obtain a transformed down-mix audio signal having a certain frequency, providing 403 a frequency-specific encoding parameter associated with the certain frequency upon the basis of the combined encoding parameter, according to the principle described herein, and synthesizing 405 the first and second audio signal at the certain frequency from the transformed down-mix audio signal and from the frequency-specific encoding parameter.
According to some implementation forms, the method depicted in Fig. 4 may be performed by the parametric decoder as shown in Fig. 2.
According to some implementation forms, the parametric decoder shown in Fig. 2 may be a parametric stereo decoder adapted for

(a) Receiving one or more combined ICC parameters, and
(b) Using for at least one parameter band an ICC parameter related to the received combined ICC parameters.

Claims

Parametric encoder for encoding a multi-channel audio signal having a first audio signal and a second audio signal, the parametric encoder having:
a transformer (101) for transforming the first audio signal into frequency domain to obtain a first transformed audio signal, and for transforming the second audio signal into frequency domain to obtain a second transformed audio signal;

a parameter generator (107) for generating a first encoding parameter, X(i), from the first transformed audio signal and from the second transformed audio signal at a first frequency band i, and for generating a second encoding parameter, X(j), from the first transformed audio signal and from the second transformed audio signal at a second frequency band j; and

a parameter combiner (109) for combining the first encoding parameter and the second encoding parameter to obtain a combined encoding parameter, X, according to the formula $X = \frac{\sum_{i \in I} g_{i} (P_{1} (i) + P_{2} (i)) X (i)}{\sum_{i \in I} g_{i} (P_{1} (i) + P_{2} (i))}$

wherein parameter I denotes a set of indices of frequency bands, parameter g_i is a weight given to frequency band i, parameter P₁(i) denotes a signal power of the first audio signal in the i-th frequency band, parameter P₂(i) denotes a signal power of the second audio signal in the i-th frequency band,

and wherein the first encoding parameter, X(i), and the second encoding parameter, X(j), is an inter-channel phase difference or an inter-channel coherence or an inter-channel intensity difference or an inter-channel level difference.
The parametric encoder of claim 1, wherein the parameter generator (107) is configured to generate the first encoding parameter and the second encoding parameter upon a basis of a multiplication of values of the first transformed audio signal and of the second transformed audio signal.
The parametric encoder of any of claims 1 to 2, wherein the parameter generator (107) is configured to generate a plurality of encoding parameters from the first transformed audio signal and from the second transformed audio signal at a plurality of frequency bands and wherein the parameter combiner is configured to combine the plurality of the encoding parameters to obtain the combined encoding parameter.
The parametric encoder of any of claims 1 to 3, being further configured to combine the first transformed audio signal and the second transformed audio signal to obtain a down-mix signal.
The parametric encoder of any of claims 1 to 4, further comprising an inverse transformer for inversely transforming a combination of the first transformed audio signal and the second transformed audio signal to obtain a down-mix audio signal.
Method for parametrically encoding a multi-channel audio signal having a first audio signal and a second audio signal, wherein the method is configured to operate a parametric encoder according to the preceding claims.