Optimized channel reduction processing of a stereophonic audio signal
The method addresses stereo to mono downmixing issues by phase aligning audio channels with IPD calculations and adaptive filtering, achieving high-quality mono conversion by reducing phase-related artifacts and maintaining audio balance.
Patent Information
- Application Number
- FR2023005109
- Authority / Receiving Office
- FR · FR
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-23
- Publication Date
- 2024-11-29
- Estimated Expiration
- Not applicable · inactive patent
Smart Images

Figure 00000000_0000_ABST
Abstract
Description
Title of the invention: Optimized channel reduction processing of a stereophonic audio signal Technical field
[0001] The present invention relates to the general field of audio signal processing. The invention relates in particular to the channel reduction processing commonly called "downmixing" of a multichannel audio signal. Of particular interest is the processing of reducing a stereophonic signal to a monophonic signal.
[0002] This type of processing finds applications generally in the field of audio technologies, and more specifically in the field of audio coding, whether at the encoding or decoding stage. Prior art
[0003] Channel reduction or downmixing consists of deducing, from a combination of the C channels of a multichannel signal x, a signal y consisting of a lower number D of channels. In practice, this consists of defining a function f (.) such that:
[0004] ' Tl ) \ , with D < C and n representing m(n) .x c (n) ! a time index of the input and output signal.
[0005] In this invention, we are interested in the particular case D=1 and C=2, we will then speak of stereo to mono downmix or simply mono downmix. In the case C=2, we speak of 2-channel content or signal which can be stereo content or binaural content, where channel 1 (Jt) is referenced as the left channel (Left), and channel 2 (x2) as the right channel (for Right). Subsequently, the stereo case will be seen as a general 2-channel signal, including the binaural case to avoid repeating the two terms, even if technically a binaural signal has specific characteristics. In the case of the signal y obtained after reduction processing, we speak of a monophonic or mono signal, which will subsequently be noted m(n), in the time domain, or M(k), in the frequency domain.
[0006] These stereo signals can come from a capture from a pair of stereo microphones or a binaural capture, or even from an artistic mix of audio tracks.
[0007] There are a number of microphone pairs that can be used to create stereo content. Among the most popular are the XY pair, which consists of two cardioid microphones with a difference in orientation angle between 90° and 135°. Another very common pair is the MS pair, for "Mid-Side", which consists of a cardioid microphone, and a second microphone called a "figure of 8" oriented at 90°. By combining these two microphones (sum / difference), we can create the left and right channels of stereo content. These pairs are said to be coincident, that is to say that the microphone capsules do not present any delay between them, the spatialization being perceived by the difference in sound intensity between the left and right channels.
[0008] Another category of so-called phase stereo pairs consists of using microphones that are distant from each other, thus creating a phase difference between the channels for sources located closer to one of the microphones. The best known is the AB pair, which uses two omnidirectional microphones spaced a few centimeters to several meters apart. Another very common pair is the ORTF pair, which uses both a phase difference through a 17cm spacing of the microphones and an amplitude difference through cardioid directivities (the microphones have an angle around 90°). The binaural pair consists of placing omnidirectional microphones in the ears of a person or an artificial head: this type of device makes it possible to natively create binaural content that can be listened to through headphones.
[0009] Stereo content also includes speech and audio signals resulting from audio mixing or post-production (e.g., channels stored on a CD or DVD, or streamed over the Internet).
[0010] There are also other methods of creating stereo content that are not reviewed here.
[0011] The simplest method for creating a downmixed signal is known as passive downmixing. It involves averaging the left and right channels of the stereo signal such that:
[0012]
[0013] or in the frequency domain: t°014i where Xj( t, k), z = 1,2 is the Fourier transform at short-term of t) (or FFT in English for “Fast Fourier Transform”), k being the frequency index and t the frame index: X^t, k) = Qe-^r, for all ke {0, ..., N-1}
[0015] Where j is the imaginary number such that j = f, t is the frame index, l is the index temporal in the current frame, Via size of the FFT, w(.) an apodization window of size L, of type sine or Hann or other, adapted to the size of the frame, and %.( t, l) = x.( t, l) ■ This definition extends to the case where N>L and x;(.) is an augmented version of x^t) with 0 (0-padding).
[0016] This very simple passive downmix method works well in situations where the right and left channels are in phase. However, in situations where the microphones are not coincident and have phase differences, such mixing generates comb filtering, which manifests itself by a coloration of the original signal. This is due to the fact that depending on the spacing of the microphones, the position of the source relative to the microphone pair and the frequency, the left and right channels of the same source can be either in phase as shown in Figure 1a, or in phase opposition as shown in Figure 1b.Also, during downmixing, the left and right signals will either add up (constructive interference) or cancel each other out (destructive interference), thus creating level variations and coloration effects of the reduced signal t ) • The coloration comes from the effect of comb filtering: the effects of constructive and destructive interference appear in different frequency bands, thus modifying the balance and therefore the timbre of the signal from the downmix compared to the original. .
[0017] To correct this intensity level defect according to the frequencies, the downmix of the e-AAC+ codec implements a level correction p(k) which ensures that the energy level of the reduced signal, in each frequency band, remains comparable to the original one: [°°181 M(k)=r(k)^?^ 100191 = IhSSm
[0020] This approach allows to compensate to a certain extent the decrease in intensity of the downmix. However, this compensation leads to overamplifications when the signals are close to phase opposition: also, in practice, the correction factor y(k) is limited to a maximum value (for example an upper limit of 2).
[0021] [Fig.7] describes a method for reducing channels from stereo to mono integrated into the public IVAS candidate codec available at:
[0022] https: / / forge.3gpp.org / rep / ivas-codec-pc / ivas-codec / -Ztree / main
[0023] The input 2-channel signal x(m), where m is the index of the interleaved samples, is deinterleaved (block 701) to find the two left and right channels. Then a frequency analysis with windowing and Fourier transform (blocks 702 and 703) is carried out to obtain the spectra of the 2 channels, and the cross-correlation based on the phase spectra is estimated (block 704) before determining the time difference between the channels (denoted here ITD even if it is formally an ICTD between 2 channels) by peak search (block 705); block 705 also provides the correlation level R* corresponding to the ITD.
[0024] The mixing factor g is then determined as follows (block 706): [°°25] nr t>q ^“î+T 2
[0026] where g represents the mixing factor and T is the interchannel time difference (mistakenly called ITD). Thus, the 2 channels are combined into a mono signal in block 706 according to this factor g, with a smoothing of the value of the factor applied sample by sample at each start of frame after block 707:
[0027] m(n) = y(n) xi(n) + (ly(n)) x2(n),
[0028] where y(n) is the factor derived from g after temporal smoothing with the gain of the previous frame.
[0029] Then, the energies of the left xi(n) and right x2(n) channels and of the mono signal m(n) are determined (block 708, 709, 710). The left and right channels are separately reinjected (added) in blocks 713 and 715 to the mono signal from block 706 according to an energy compensation determined in block 711 with respective scale factors defined in blocks 712 and 713.
[0030] These methods, in addition to the coloration of the signal from the downmix, linked to the imperfectly corrected comb filtering, suffer from an excess of reverberation linked to the summation of signals when the input channels are out of phase, and from a "click" type artifact when the reinjection of channels is carried out punctually in a single frame. Compensating for an identical delay for all frequencies is not realistic, this delay being linked to the different sources making up the content. In practice, this leads, in addition to additional coloration, to a potential loss of intelligibility of the sources, a defect that level compensation cannot correct.
[0031] Other downmixing approaches seek to avoid the aforementioned defects. The principle of these methods is to re-phase the left and right signals before their summation. In the document entitled "A stereo to mono downmixing scheme for MPEG-4 parametric stereo encoder" by Samsudin, E. Kurniawati, N. Boon Poh, F. Sattar, S. George, in Proc. ICASSP, 2006, a method is proposed which consists of applying, in the frequency domain, a phase shift q> on one of the channels, generally the right, in order to re-phase it with the left channel, such that:
[0032] M j = x,(k)+x2(k)^
[0033] The ideal phase shift is given by what is called here the IPD for “Inter-channel Phase Difference” in English:
[0034] IPD(k) = 4^1(^2(^) )
[0035] where X" is the conjugate of X and L indicates the phase of the complex operand.
[0036] It is calculated in the frequency domain: this allows the com to be put back in phase spectral characteristics of different sources, a source with its own IPD being able to be dominant in a frequency band k, while another source with a different IPD can be dominant in another frequency k'.
[0037] In the case of the MPEG4 codec described in the Samsudin document cited above, the IPD applied is that estimated by averaging over a Bark band and not the IPD calculated for each frequency band, the latter proving to be particularly noisy and variable from one frame to another. In addition, this method takes the left channel as phase reference, and if the phase of this channel is poorly conditioned, the downmix has degraded quality.
[0038] In the published patent application WO2017103418, a downmix method is proposed which makes a mixture between the method proposed by Samsudin and the passive downmix. In particular, it proposes an ISD indicator for "Inter-Spectral Distance" which makes it possible to choose, frequency band by frequency band, which is the most suitable method for rephasing the signals:
[0039]
[0040] This indicator makes it possible to measure whether the signals are rather in phase (ISD < 1), i.e. with a phase difference or IPD in the interval |^ _ 2. j or rather in phase opposition (ISD > 1), i.e. with an IPD in the interval j. When they are in phase opposition, a large ISD value is synonymous with a comparable level between the left channel and the right channel and the phase adjustment is important to avoid the effects of coloration and loss of level. Conversely, an ISD value close to 1 is representative of one channel being predominant compared to the other; in the latter case, the phase adjustment is not important, the downmix being approximately equal to the predominant signal. It is, however, important to avoid modifying the phase of the right channel if it is predominant. Also, in this patent application, the following downmix type selection mechanism is proposed: [00411 if ISD(k)> 1.3 I M.(k) = if 1SDy^ < J 3
[0042] The passive downmix proposed in the aforementioned patent application, for low ISD situations, sometimes introduces increased reverberation.
[0043] Furthermore, it appears that in certain situations of signals which are rather in phase, i.e. when ISD(k) < 1.3, and in particular when the channels are highly unbalanced (source predominantly on the left or on the right), there is a “distortion shifting” phenomenon which can be troublesome, it can therefore be important in this case to keep this signal intact without applying a phase shift.
[0044] On the other hand, the method proposed in the aforementioned patent application requires applying processing in the frequency domain, the reconstruction of the reduced signal being carried out from the short-term inverse Fourier transform (TFCT or STFT in English) of the signal M(t,k). This type of filtering generates a circular convolution which creates audible artifacts: echo, pre-echo, crackles. To mask these artifacts, it is possible to use an implementation with an overlap of frames associated with a suitable windowing (window for analysis and / or synthesis) guaranteeing a reconstruction of the filtered signal. In general, this overlap is not compatible with the operation of audio coders which work with adjacent frames, i.e.without overlap; moreover, this type of reconstruction induces a delay, generally of half a frame which is the classic overlap, which is not acceptable for a coder which has strong constraints in terms of latency.
[0045] Another implementation of the Overlap-and-Add (OLA) type is also possible: this method requires a 0-padding step on the signals and filters to avoid circular convolution. However, tests show that with such filters whose phase varies very quickly from one frame to another, an OLA type approach does not allow for completely masking the artifacts at the transition of certain frames, the artifacts then remaining audible.
[0046] There are also state-of-the-art methods where the downmix is obtained by switching, for example in the frequency domain for the published patent application WO2017103418, different downmix methods. In this case it is important to ensure that the switching is done "seamlessly", that is to say without discontinuities or differences in levels between methods to avoid artifacts. Presentation of the invention
[0047] The invention improves the state of the art.
[0048] To this end, the invention aims at a method for processing channel reduction of a stereophonic signal to obtain a monophonic signal comprising, for a current frame: - a first step of selecting a channel reduction treatment between two methods using filtering of the stereophonic signal and rephasing of the signals, the selection being carried out according to a phase indicator, representative of a measurement of the degree of phase opposition between the frequency components of the channels of the stereo signal,
[0049] - a second step of selection between one of the two methods, selected at the outcome of the first selection step and a channel reduction processing method without rephasing, the selection being carried out according to the value of a ratio of energy between the channels of the stereo signal.
[0050] These selection steps thus make it possible to determine the downmix best suited to the characteristics of the stereophonic signal.
[0051] In one embodiment, the energy ratio between the channels of the stereo signal is calculated per frequency band, the second selection step being carried out per frequency band.
[0052] Thus, only the frequency bands concerned, whose energy ratio exceeds a threshold, apply a downmix without re-phasing for finer adaptation.
[0053] In a particular embodiment, the channel reduction processing without rephasing is selected for the frequency bands concerned only in the case where the number of frequency bands concerned exceeds a threshold.
[0054] This thus makes it possible to avoid excessive fluctuations from one frame to another.
[0055] In one embodiment, a third selection step is performed between one of the two methods selected at the end of the second selection step and a third channel reduction processing method not using filtering of the stereophonic signal, said selection being implemented as a function of:
[0056] - a presence or absence of at least one transition in the stereophonic signal, or
[0057] - a quality level of the filtering of the stereophonic signal used during the implementation implementation of a treatment using stereophonic signal filtering, or
[0058] - an indication of the presence of a binaural signal.
[0059] This embodiment allows:
[0060] - when one or more transitions are detected, to obtain a better rendering on the or the transition(s) of the signal by selecting the channel reduction processing which does not use filtering, and which, therefore, preserves the nature of the stereophonic signal,
[0061] - when no transition is detected, to favor the channel reduction processing channels that use filtering, better suited to stereo signal components that are not in phase.
[0062] In one embodiment, the detection of said at least one transition is implemented before applying, to said stereophonic signal, one or the other of the two channel reduction treatments.
[0063] This embodiment makes it possible, even before applying channel reduction processing, to deduce a priori whether at least one transition is present or not in the stereophonic signal, without applying any additional particular processing to this signal.
[0064] In one embodiment, the detection of said at least one transition is implemented as follows:
[0065] - calculation of the energy of the stereophonic signal, called input energy, before application filtering,
[0066] - calculation of the energy of the monophonic signal, called output energy, after application filtering,
[0067] - calculation of a ratio between the input energy and the output energy,
[0068] - comparison of said ratio with an energy threshold, and
[0069] - if said ratio is lower, respectively higher, than said energy threshold, the channel reduction processing that does not use stereophonic signal filtering, respectively that uses stereophonic signal filtering, is selected.
[0070] In one embodiment, said filtering quality level is compared to a filtering quality threshold, and if said filtering quality level is lower, respectively higher, than said filtering quality threshold, the channel reduction processing which does not use stereophonic signal filtering, respectively which uses stereophonic signal filtering, is selected.
[0071] In one embodiment, an indicator is generated representative of the result of said detection, of the comparison of said ratio with an energy threshold, or of the comparison of the filtering quality level with a filtering quality threshold.
[0072] In one embodiment, the method comprises a preliminary phase of adapting the channel reduction processing filters before application to the stereo signal by the following steps: - obtaining the impulse responses of the filters corresponding to the channel reduction processing; - truncation of part of the impulse responses; - weighting of the remaining part by applying a weighting window; - normalization of the resulting impulse responses from windowing to obtain the appropriate filters, to be applied to a stereo signal frame.
[0073] In a particular embodiment, the truncation consists of retaining a causal part and a non-causal part of the impulse responses.
[0074] The invention relates to a channel reduction processing device comprising a processing circuit for implementing the steps of the channel reduction processing method as described previously.
[0075] The invention also relates to a computer program comprising instructions for implementing the channel reduction processing method according to the invention, according to any one of the particular embodiments described above, when said program is executed by a processor.
[0076] Such instructions may be stored durably in a non-transitory memory medium of the channel reduction processing device implementing the channel reduction processing method according to the invention.
[0077] This program can use any programming language, and be under the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any other desirable form.
[0078] The invention also relates to a recording medium or information medium readable by a computer, and comprising instructions of a computer program as mentioned above.
[0079] The recording medium may be any entity or device capable of storing the program. For example, the medium may comprise a storage means, such as a ROM, for example a CD ROM or a microelectronic circuit ROM, or a magnetic recording means, for example a mobile medium, a hard disk or an SSD.
[0080] On the other hand, the recording medium may be a transmissible medium such as an electrical or optical signal, which may be conveyed via an electrical or optical cable, by radio or by other means, so that the computer program it contains is remotely executable. The program according to the invention may in particular be downloaded over a network, for example an Internet-type network.
[0081] Alternatively, the recording medium may be an integrated circuit in which the program is incorporated, the circuit being adapted to execute or to be used in the execution of the aforementioned channel reduction processing method.
[0082] According to an exemplary embodiment, the present technique is implemented by means of software and / or hardware components. In this regard, the term "device" or "module" may correspond in this document to a software component, a hardware component or a set of hardware and software components. Brief description of the drawings
[0083] Other characteristics and advantages of the invention will appear more clearly on reading the following description of particular embodiments, given as simple illustrative and non-limiting examples, and the appended drawings, among which:
[0084] [Fig. 1a] illustrates channels of a stereophonic signal, in phase as previously described;
[0085] [Fig.lb] illustrates channels of a stereophonic signal, in phase opposition as previously described;
[0086] [Fig.2] illustrates in block diagram form a channel reduction processing chain in one embodiment of the invention;
[0087] [Fig.3] illustrates an embodiment of a filter calculation for channel reduction processing, as well as a channel reduction processing selection;
[0088] [Fig.4] illustrates an embodiment of an adaptation phase of a channel reduction processing filter;
[0089] [Fig.5] illustrates an embodiment of a transition of the application of the treatment channel reduction during a transition between two stereo signal frames;
[0090] [Fig.6a] illustrates another embodiment of a filter calculation for processing channel reduction, as well as a selection of channel reduction processing;
[0091] [Fig.6b] illustrates another embodiment of a filter calculation for processing channel reduction, as well as a selection of channel reduction processing;
[0092] [Fig.7] illustrates an existing channel reduction processing method described above cededly;
[0093] [Fig.8] illustrates an example of structural embodiment of a device for processing channel reduction according to one embodiment of the invention. Description of the embodiments
[0094] The invention will now be described below with reference to Figures 2 to 8. For this purpose, the following symbols will be mainly used in the description which follows. INDICATIONS
[0095] k frequency index
[0096] 1 time index of a signal (in the current frame)
[0097] n time index of a signal
[0098] t frame number CONSTANTS The number of samples in the current frame
[0099] R filter delay
[0100] N length of FFT
[0101] C number of input channels
[0102] D number of output channels
[0103] P filter length
[0104] Q crossfade length SIGNALS
[0105] Xi(n) input channels
[0106] m(n) downmix (in time)
[0107] m(t,l) downmix in the current frame (in time)
[0108] M(t,k) downmix in the current frame (in frequency)
[0109] [Fig.2] shows an example of a processing chain for a stereophonic audio signal, the processing comprising a channel reduction process known as “downmix”.
[0110] At the input of this processing chain, a stereophonic signal x, composed of two channels (xi(n) and x2(n)) also called left channel and right channel (respectively), is firstly divided into 201 frames of L samples (xi(t,Z) and x2(t,Z), t being the frame index and 1 the sample index. The block 202 applies windowing and an FFT (Fast Fourier Transform) to obtain signals in the frequency domain (X,(7A) and X2(t,k), k being the frequency index). In 203, a selection and determination of a channel reduction filter to be applied to a signal frame are then carried out. This step will be described with reference to [Fig.3]. The filter thus determined is conditioned (or adapted) in 204 to make it causal or partly causal and to optimize it in order to reduce the processing complexity. In an exemplary embodiment, the windowing and the FFT are similar to steps 702 and 703 of the [Fig.7]. [YES]
[0112]
[0113]
[0114]
[0115]
[0116]
[0117]
[0118]
[0119]
[0120]
[0121] This adaptation phase is described with reference to [Fig.4]. Once the filter(s) have been determined and adapted for the current frame, they are applied in 205 to this current frame. In order to avoid audible artifacts due to a change of filters between two frames, a crossfade is performed taking into account counts the filters of the previous frame ( t - 1, l ) when downmixing the U) previous frame is of the same type (T1 / T2 defined later). This step will be described with reference to [Fig.5]. [Fig. 3] describes a detailed embodiment of block 203 of [Fig. 2]. In this processing block, a first or second channel reduction processing to be applied to a current frame of the stereophonic signal is selected and determined in a first selection step. A first method of channel reduction processing Tl is defined here by shifting the left and right channels of the stereo signal simultaneously by an angle defined by a quarter of the phase difference (IPD) determined between the two channels of the stereo signal. This gives a downmix signal, as follows: k) = This method avoids excess reverberation type degradation generated by a classic downmix (average of the two signals) or comb filtering type in the event of a time lag between the two channels. This downmix thus defined can be seen, in the frequency domain, as a filtering according to the following equation: M(t, kjX^t, k) + k)X2(t, k) with H^k) = and = In another embodiment, the first reduction processing method Tl channels is defined by shifting the left and right channels of the stereo signal simultaneously by an angle defined by half the phase difference (IPD) determined between the two channels of the stereo signal.
[0122] We then have:
[0123] =
[0124] Or equivalently
[0125] M(t,k) = H^t,k)X1(kk)+H2(t, k)X2(t,k) mi .IPVxt.D, , JPDÎrS}
[0126] ^ecH^tky = e^etH^tky=:^.
[0127] A second T2 downmix method is for example defined as that proposed in the previously cited Samsudin document.
[0128] In this second method, we then have — 1 and H2( t. k) =
[0129] As described further, in one embodiment, the impulse responses corresponding, h^t, l)' are in fact shifted in time to keep the causal and anticausal parts. This shift can in variants be integrated directly into the definition of kk)-
[0130] Furthermore, in variants, it will be possible to equivalently remove the factor U2 from the definition of H^t, k) and integrate this factor U2 when mixing the channels treated in 504 and 505 with reference to [Fig.5] described later.
[0131] In practice, the phase difference IPD or IPD / 4 or IPD / 2 does not need to be determined. In an efficient implementation, the filters H t. k) can be determined by normalizing the (complex) frequency lines - with or without smoothing - by their modulus (in the sense of complex numbers).
[0132] The channels of the stereo signal, in the frequency domain (X[(t,k) and X2(t,k)), are used on the one hand to calculate a phase indicator ISD in 301, representative of a measurement of the degree of phase opposition between the channels of the stereo signal, and, on the other hand, the phase difference between the channels (IPD) in 303.
[0133] The phase indicator is for example defined by the ISD indicator for “Inter-Spectral Distance” as defined above.
[0134] In the embodiment described herein, the ISD is determined per stereophonic signal frame so that the decision to select a first channel reduction process (first downmix) or a second can be made on a frame-by-frame basis.
[0135] This decision by frame makes it possible to avoid significant phase shifts between frequencies and other audible artifacts which can appear when the decision varies from one frequency to another as in the patent application cited above.
[0136] To apply a decision per frame, a decision criterion according to the ISD phase indicator is defined according to the following equation at 302:
[0137]
[0138]
[0139]
[0140]
[0141]
[0142]
[0143]
[0144]
[0145]
[0146]
[0147]
[0148]
[0149]
[0150] r cr» r \ total number of frequency lines where ISD{ t,k )>l .3 ' ' J total number of frequency lines This criterion measures the percentage of frequency lines of index k for which the criterion is favorable to downmixing with re-phasing, i.e. when ISD(t, k) > 1.3- The decision is made to apply the first downmix as defined above, with rephasing, when ISD(t) exceeds a certain threshold l§Dth, which is set for example at 70%. In another embodiment, the ISD indicator can be determined on a predefined frequency band: ISD(t) = .....j.....lSD(t, k) where for example kmin = 2 and kmax is the number of the frequency line corresponding to a maximum frequency, for example 4kHz. This criterion corresponds in this case to an average of the ISD value per frequency line. In this variant, the threshold corresponds not to a percentage but to an ISD value, and we can set for example ISDth = 1.5- We then have in 305 the following decision of downmix selection between T1 and T2: ISD(t) <lSD,h : k) H2(t,k)=^- iSD(t) > iSD* ; H^k,f)=^^- In 303, the phase difference IPD is obtained between the two channels of the stereo signal P^IPD(t,k) = ^X}(t,k)^(t,k)) where z indicates the phase of the complex operand. However, the frame decision may sometimes be insufficient because 1TPD, as calculated according to the formula above, varies very rapidly over time. Also, filters Hk) can vary very quickly from one frame to another, which causes discontinuities that are sometimes audible during the transition from one frame to another. To limit this effect, we can initially calculate the IPD on frequency sub-bands rather than in each band independently, by averaging each channel over wider sub-bands. We can choose a number B of sub-bands of width equal to BK — K ! B, where B is an integer greater than or equal to 1. To calculate the IPD in each sub-band b, we first calculate the spectra Xi(t, b) in each sub-band by averaging: The average spectra of the sub-bands corresponding to the zero frequency (b=0) and the Fe / 2 frequency (b — B / 2) are treated separately by direct deduction of the spectra in these frequency bands:
[0151] ^-(^0)=^(^0)
[0152] X i (t,B / 2)=X i (t,Kf2)
[0153] The number B of sub-bands is to be determined experimentally. It will be necessary to maintain a sufficient frequency resolution to be able to apply a phase adjustment adapted to the different sources: in practice, a frequency resolution around a 100 Hertz seems a good sub-band width. For example, for a stereo signal sampled at 32kHz, with an FTT size of 640, this corresponds to a number B = 320 sub-bands with BK = 2 frequency bands in each sub-band.
[0154] In variants, it will be possible to use a division into sub-bands with a non-uniform width, for example according to the Bark scale.
[0155] Then, the IPD is calculated in the same way as previously, i.e.: [0!56] IPD(b) = ^X^X^b) )
[0157] In the following, the index k will indicate indifferently either a frequency band or a frequency sub-band, the case BK=1 being equivalent to the processing in each frequency band separately.
[0158] Still to limit the effect induced by rapid variations of the IPD, we can use a smoothed version of the IPD over time to calculate the filters H^t, k )■ One way to smooth the IPD is to apply a low-pass filtering of type IIR of order 1, in 304, such that:
[0159] IPD(pk) a(k)IPD(t, k) + k)
[0160] The forgetting coefficient a(k} is chosen ad-hoc. It may be interesting to choose a high forgetting coefficient in the low frequencies, typically below 5kHz, because this allows phase coherence to be maintained in the low part of the spectrum from one frame to another, the spectrum of speech signals being stationary, particularly in its low part and the particular ear sensitive to the phase in this part of the spectrum. Conversely, it is interesting to choose a low forgetting coefficient in the high frequencies: in the high frequencies, typically above 5kHz, the phase continuity of speech signals is less marked and the ear is less sensitive to phase discontinuities in this part of the spectrum.
[0161] We can choose a forgetting coefficient of the following form: bk[mv &mitp k ^high amin + , ^!ow khigh
[0163] The values amin^ khiah and kiow are determined experimentally. In the a(k) = case of adjacent frames of length L=20ms, we can for example choose the following set of parameters:
[0164] amax — 0.94, a„„„ = 0.86, klow — 0, corresponds to 5kHz
[0165] In variants, we will force the very low frequencies and the very high frequencies to remain intact. Thus in variants, we force H^t, k) = H2(l £)=4 For the first lines k = 0 or k = 0, 1 as well as for the lines of index k > kmax where k corresponds to 8 kHz.
[0166] Downmixes with re-phasing, T1 or T2, can sometimes significantly modify the timbre and cause degradations in the quality of the signal. These timbre modifications are particularly audible when a signal is predominant on one or the other channel, that is to say when the ILD for "Interchannel Level Difference" which measures the energy ratio between the left channel and the right channel is close to 0 or much higher than 1. This is for example the case of a signal predominant on the right channel to which the T1 re-phasing mode is applied by 1TPD. To limit these timbre modifications, it is interesting when the energies of the left and right channels are very different to avoid re-phasing by applying a conventional downmix.
[0167] In a first implementation, an independent decision can be made in each frequency band, which makes it possible to process different sources independently, one of which can be predominant on a channel and does not require rephasing, while the others are distributed over the left and right channels and require rephasing. In this case, a passive downmix will be forced only in the frequency bands which present a strong energy disparity:
[0168]
[0169] k) = H2{kk) =4, V k, if ILD(t, k) > a!LD or ILD(t, k) < djLD Or 11 ni \ „ I^jÙCI2 and ajLD is a threshold the ILD defined experimentally. In In practice, we can choose a value around 100 (20dB).
[0170] Thus the ILD is calculated in 306 in order to implement a second selection step in 307 and thus determine the downmix processing to be applied with or without re-phasing T1, T'1, T2 or T'2 and according to the frequency bands.
[0171] The decision to impose a passive downmix independently in each sub-band can sometimes cause phase fluctuations from one frame to another: this is particularly audible when few frequency bands are concerned. To avoid these fluctuations, we can limit the application of the passive downmix in the bands concerned only if we observe enough bands concerned, synonymous with the effective presence of a source on the left or right. In practice, we can calculate the percentage of bands concerned ILDprc(t}, and limit the application of the passive downmix in the bands concerned defined previously if this percentage is higher
[0172]
[0173]
[0174]
[0175]
[0176]
[0177]
[0178]
[0179]
[0180]
[0181]
[0182]
[0183] at a predefined threshold aiLDprc, or: jT r\ f \ _ number of bands where lLD^tk}>aILD ouILD(tJi) <l / alLD pic \ ) — nombre total débandés de fréquence And aTLDPrc CS( a scuj| determined experimentally. In practice, a threshold of 30% is a good compromise. Figure 4 now details block 204 of Figure 2. The filters t, k), as for example defined in the downmix methods T1 and T2 defined above, are reconditioned or further adapted to avoid circular convolution. In a first step, in 401, we calculate the impulse responses hj^ fl) of the filters Hz(t, k), such that: h^fl) =FFrl[Hi(î, k)}, 1,2 where the filters fl) are of size Bp = B. In the preferred embodiment, we have L = 20 ms, or K = 320 samples at 16 kHz, 640 samples at 32 kHz and 960 samples at 48 kHz, BK = 2, or B = 160 samples at 16 kHz, 320 samples at 32 kHz and 480 samples at 48 kHz. Impulse responses thus defined are classically non-causal; they correspond to filters with a finite impulse response. A classic technique for making them causal is to perform a rotation of B / 2 samples to center the response at B / 2. The disadvantage is that the filter thus reconstructed then has a latency of B / 2. In the preferred embodiment, a truncation of the impulse response will be carried out: h^fl-M), M< l< P h'ifB-M + l) 0< KM This (circular) time shift operation can also be implemented by applying an appropriate phase term to Ht, k ) before inverse FFT, as known to those skilled in the art. In this embodiment, the resulting filter has a finite impulse response with a delay of R samples. This delay allows the truncated filter to be properly conditioned, either in terms of gain by minimizing the influence of truncation on the unity gain of the phase-shifting filter, or in phase by minimizing the deviation to 1TPD. This delay R can be adapted according to the implementation constraints: in practice, a delay around one millisecond is sufficient to properly condition the truncated filter. In an extreme variant, the truncation is asymmetrical to avoid this latency. We choose to truncate the filter asymmetrically, at 402, by retaining only a portion of the first half of the impulse response fl) in the following manner:
[0184] h^tj) = Z)<0< 1 <P<b / 2
[0185] This filter has the advantage of generally having a maximum (in absolute value) on its first sample, which guarantees processing without additional delay. In terms of frequency response, it has a phase roughly equal to the optimal rephasing filter h^t, l}- It may be possible to compensate for the fact that the anticausal part of the impulse response is suppressed by additionally applying a factor of 2 to the values of Z) for / > 0. In this variant, the resulting filter has a finite impulse response with virtually no delay.
[0186] The choice of P depends on different criteria: a value of P close to N / 2 guarantees a near-optimal phase re-phasing, while a lower value limits the computing power required for filtering. A reduced value of P also smooths the phase of the filter. This has an advantage with filters defined by T2 processing as defined above where the phase varies very quickly from one frequency to another: these variations can result in a very high group delay perceptible to the human ear. A low value of P allows all or part of these artifacts to be eliminated.
[0187] The filter thus truncated t, l), in particular when P <N / 2, voire P«:N / 2, présente une queue de réponse impulsionnelle qui ne tend pas vers 0. Cela crée des discontinuités audibles de type cliquetis à la transition entre 2 trames. Aussi, pour éviter ces artefacts audibles, on pondère h^t, Z) avec une fenêtre w(Z), en 403, tel que :
[0188] Z)
[0189] where the coefficients of w(Z) tend towards 0 when 1 tends towards 0 or P and is equal to 1 at l=R. This makes it possible to reduce the energy of the samples at the end of the filter and to avoid discontinuities during the transition between filters of consecutive frames.
[0190] In the preferred embodiment, where the truncation preserves an anticausal part, for example, U2 triangular windows are applied to the causal and anticausal parts.
[0191] |,0 <l<R w(Z) = , 1-^,RS1 <P
[0192] or any other window of the Hann type for example*
[0193] In the variant where the truncation is asymmetrical, this window can be a triangular half-window, or even a Hann half-window:
[0194] w(l) = 1-1 / P, OR COsÇ / tT / 2P)
[0195] Due to its truncation and windowing, the filters h;(t, Z) have a reduced energy compared to that of the optimal filter h^t, Z): the latter, as a pure phase shifter filter, has a norm 1 / 2 by construction. Also, we can proceed to a normalization of the energy of the filter thus truncated and windowed, in 404, such that:
[0196] ' 2||Â,(^)||
[0197] Where || x(.) || is the 2-norm of a vector.
[0198] In variants, another method can be used in 204 and the method of [Fig.4] can be replaced. For example, the impulse response can be determined directly by least squares minimization and matrix inversion using the (complex) Levinson algorithm as described in section 2 of the article by Mathias C. Lang, Design of nonlinear phase FIR digital filters using quadratic problems, Proc. ICASSP, 1997. An implementation of this method can be found, for example, in the thesis by Mathias C. Lang, Algorithms for the Constrained Design of Digital Filters with Arbitrary Magnitude and Phase Responses, June 1999 (see routine levin.m in Appendix B and Islevin.m in section 2.1.4). In this case, the inverse FFT, truncation and weighting are not necessary; This alternative method directly estimates a finite impulse response of length P (without a minimum phase guarantee). However, a normalization can be applied in the form
[0199] According to the invention, at 405, a quality level of the filtering that has been implemented is determined. In the preferred embodiment with truncation with a causal part, the level of |h.( / ) | for 1 = R can be determined. When this value is lower than a predetermined threshold (for example 0.25), this indicates a low quality criterion, and conversely, when this value is higher than a predetermined threshold (for example 0.25), this indicates a good quality criterion.
[0200] In variants, it is possible to determine for this purpose
[0201] - an energy E40i of the filter before truncation 402 is calculated,
[0202] - an energy E403 of the temporal filter after windowing 403 is calculated,
[0203] - a ratio between E40i and E403 is calculated.
[0204] A filtering quality can also be determined by comparing an energy E40i and an energy E404 of the temporal filter after normalization.
[0205] As non-exhaustive alternatives, the determination of the filtering quality level can be implemented:
[0206] - directly on the filter H^k), and / or
[0207] - a frequency version of this same filter before normalization Hi( î. k} or after normalization k)-
[0208] -etc...
[0209] In other variants, we can calculate: 102101
[0211]
[0212]
[0213]
[0214]
[0215]
[0216]
[0217]
[0218]
[0219]
[0220]
[0221]
[0222]
[0223]
[0224]
[0225]
[0226]
[0227]
[0228] This criterion can be standardized by: LA:(M) 2 Another quality variant may be interested in measuring the amount of energy lost during truncation to make the filter causal. In this variant, we can calculate E40i as the "anticausal" energy of the original filter: £401( 0 = The energy E403 (resp. E404) can be calculated as the energy of the synthesized filter, before normalization (resp. after normalization): £»3(0 =0? U 0 ^404(0 0.9 To avoid the bias linked to the size of the window, we can also calculate the energy of the anticausal part on the same number of samples, with windowing by the window w ( / ) or not: ^401(0 = i) £4O1O) = ^^(^^ + ^)7^ z) In variants, we can calculate the frequency response H^t, k) of the filter obtained after truncation (at the output of 403) and compare with the target response Hf (t, k) to define the quality level, for example: This criterion can be standardized by: Ll«,U*)l 2 Another variant of the quality criterion could be to compare the phase coherence of the filters, given that the objective is to re-phase the signals. Thus, we could calculate the coherence between the two filters: r / A - _L ly A high coherence (i.e. close to 1) indicates that the synthesized filter is close to the ideal filter and therefore a high quality factor, a low value (close to 0) indicating on the contrary a low quality criterion. [Fig.5] now describes in detail block 205 of [Fig.2], where an OLS (Overlap-and-Save) implementation of the filtering is carried out in the time or frequency domain. The choice of domain will be made according to the complexity which depends on the size of the filter relative to that of the frame to be processed.
[0229] Here we present an OLS type implementation in the time domain, with adjacent frames of size L.
[0230] First, we concatenate, in 501, the current frame with the Pl samples of the previous frame (Save principle):
[0231] = [x^tL-P+2), ...to;((z+1)L-1)]
[0232] Each channel is filtered by its phase shifter filter at 503:
[0233] OiKL
[0234] The mono downmix is then created by adding the two channels back into phase at 505
[0235] m[t, l) = yJ)0 <l<L
[0236] In practice, the filters can be very different between two successive frames, for example when a new source appears. In this case, the transition between l) and m( t, l) can cause differences in the form of audible artifacts. To avoid these artifacts, a cross-fade step can be applied. To achieve this cross-fade, the filters t - 1, l ) are applied to the current frame over a number of samples Q: I®2'7' 3^0, / )=01,tl-lp'Wt.l + Pp), Oil
[0238] This is achieved by block 502 of [Fig.5].
[0239] The number of samples required for the cross-fade will have to be determined experimentally, depending on the size of the filters, the sampling frequency, etc.
[0240] We then create the mono downmix rnt_| ( / , l ) in 504, from filters of the previous frame:
[0241] nV1( / ,Z)=yv / 7J)+^ 0 < / <Ô
[0242] The signal from the downmix m(t, l) is created at 506, from a cross-fade according to the following equation:
[0243] (^( / )^( / , / ) + (1^( / )). / 7 / ( / , / ), 0 < / <g m(t, l), Q <l<L
[0244] Where fi (l) is a monotonically decreasing function, with (0) = and fi( Q - 1 ) = g with e a value close to 0 or zero. Typically, we can choose a half-Hann window: [°245] fi(l) =cos(^ ),() < / < Q
[0246] It should be noted that, in an integration with other downmix methods, the aforementioned downmix methods T1 and T2 cannot always be represented by a filtering operation. This is the case of method T3 of [Fig.7] or any other method whose gain depends on the instantaneous level of the signal.
[0247] In the preferred embodiment, the T3 method of [Fig.7] is modified to add a time offset (delay) to synchronize the output downmix with the downmix of the T1 / T2 methods. In addition, energy compensation is applied to the T3 method in the same way as to the T1 / T2 methods to ensure level consistency and allow for better transition between methods.
[0248] According to the invention, in the case where a selection of the downmix methods T1 and T2 or of the downmix T3 is implemented, as will be described in the remainder of the description, a scaling step 5040 and 5050 of each of the signals (t, Z) and m^t, Z) respectively is implemented. For this purpose, the signal m'(n) obtained at the output of the downmix T3 of FIG. 7 having a level N3, the level NI, respectively N2, of each of the signals m{_ j ( f, Z ) and m (t, Z ) on which the downmix T1, respectively T2, has been applied, is adjusted to the level N3, so as to be equal or similar to this level N3. Instead of being applied to each of the signals m^t, Z) and m( t, Z), such scaling can also be carried out on the signal Z) •
[0249] [Fig.6a] now details a channel reduction processing method according to the invention, in which one of the downmix methods T1 or T2 (or T'1 / T'2 according to the second selection step) chosen at the end of the selection method of [Fig.3] is implemented at 610, and another downmix method is implemented at 611, the gain of which depends on the instantaneous level of the signal, such as for example the method T3 of [Fig.7].
[0250] In this case, one solution is to calculate the downmixes of each method for the current frame of time index / and to operate a transition at 612 from one downmix to the other by cross-fade, which generates a monophonic signal m(n). More precisely, in the case of a transition from a downmix T1 (or T2) to the downmix T3, the downmix can be carried out between the signal m(t, Z) obtained at the output of T1 (or T2) and the signal / ) obtained at the output of T3. The cross-fade can also be carried out between the output of Tl (or T2) and the signal d(n) as described in [Fig.7]. In the latter case, the energy compensation of the T3 method (adapt_gain in [Fig.7]) would apply permanently to all methods. The decision to switch from a Tl or T2 downmix to a T3 downmix (which is a third selection step) could be motivated by a criterion such as 1TSD defined above, or by other criteria, for example Par3 parameters already calculated by the T3 method, such as 1TTD, by choosing to apply the T3 downmix by default when 1TTD is very large (for example ITD > 600 microsec in absolute value). The ILD(t) criterion could also be used when one channel is much higher than the other (case of a zero signal on the left or right for example): the downmix without re-energization would be preferred. phase to avoid any unstamping. Other criteria or combinations of criteria can also be used.
[0251] In particular, the third step of selecting the downmix T1 (or T2) or T3 is implemented as a function of:
[0252] - a presence or absence of at least one transition in the stereophonic signal, or
[0253] - a quality level of the filtering of the stereophonic signal used during the implementation work of the Tl (or T2) downmix, or
[0254] - an indication of the presence of a binaural signal. This latter indication is an in external training Fbinaurais (e.g. defined as a binary value: l=binaural, 0 = other) which explicitly indicates that the input signal is binaural.
[0255] In the case of a detection 613 of the presence or absence of at least one transition in the stereophonic signal, according to a first embodiment, a transient detection is applied separately to each of the input signals Xi(n) and x2(n), or on the current frame Xi( / ,Z) and x2( / ,Z). Transient detection being a classic problem in audio coding, we take the example of the module already implemented in the EVS codec and described in the 3GPP TS 26.445 standard clause 5.1.8. For this purpose, each of the input signals xi(n) and x2(n), or the current frame xi(t,Z) and x2(t,Z) is divided into sub-blocks. The energy of each sub-block is calculated, then smoothing is possibly implemented. The energy obtained is then compared to an energy threshold th 6i. If this obtained energy is less than th6i, the presence of a transition is considered as not detected. If this obtained energy is greater than th6i, the presence of a transition is considered as detected.For this purpose, a parameter Par 1 (if T1 was chosen after selecting [Fig.3]) or Par2 (if T2 was chosen after selecting [Fig.3]) is set: .
[0256] - to a first value, for example 0, to indicate that the presence of a transition was not detected,
[0257] - to a second value, for example 1, to indicate that the presence of a transition has been detected.
[0258] Such a Parl / Par2 parameter is transmitted, as a decision criterion, to the cross-fading block 612.
[0259] In the case where the selection of the downmix T1 (or T2) or the downmix T3 is based on a quality level of the filtering of the stereophonic signal used during the implementation of the downmix T1 (or T2), with reference to [Fig.4], the following steps are implemented:
[0260] - calculation of the energy E6i of the signal t, l),
[0261] - calculation of the energy E62 of the signal 7(.( / , Q,
[0262] - calculation of a ratio between E6i and E62,
[0263] - comparison of said ratio with an energy threshold SE.
[0264] If the ratio between E6i and E62 is less than SE, the presence of a transition is considered as not detected. If the ratio between E6i and E62 is greater than SE, the presence of a transition is considered as detected. For this purpose, a parameter Par 1 (if T1 was chosen at the end of the selection of [Fig.3]) or Par2 (if T2 was chosen at the end of the selection of [Fig.3]) is set:
[0265] - to a first value, for example 0, to indicate that the presence of a transition was not detected,
[0266] - to a second value, for example 1, to indicate that the presence of a transition has been detected.
[0267] Such a Parl / Par2 parameter is transmitted, as a decision criterion, to the cross-fading block 612.
[0268] Thus, according to the invention, when a transient is detected in the current frame in one or both channels, the downmix T3 is applied to the current frame rather than the downmix T1 (or T2). Indeed, the downmixes T1 and T2 use filtering by an impulse response which sometimes tends to spread the envelope of the signal; this problem is more pronounced on transient sounds such as castanets, scissors clicking on a binaural recording simulating a haircut at a hairdresser's, etc.
[0269] When external information Fbinaurais explicitly indicates that the 2-channel input signal is binaural, i.e. Fbinaurais =1, a particular downmix can be chosen according to the invention, for example the downmix T3.
[0270] In order to limit potential artifacts due to crossfading (block 612) between downmix methods, the invention may provide for applying, at 614, a scaling of the signal at the output of T1 (or T2). According to the invention, such scaling 614 implements the following steps:
[0271] - determine the energy E' of the downmix signal at the output of Tl (or T2), so as to preferential with smoothing, by performing the same steps as steps 708, 709, 710 of [Fig.7],
[0272] - determine the scale factor:
[0273] g' = lEi+E2 N 2 E'
[0274] - apply the factor g' to said downmix signal using the principle of func operation (with smoothing at the start of the frame) of blocks 712 and 714 of [Fig.7].
[0275] Unlike the downmix in [Fig.7], here we apply a direct compensation of the downmix level at the output of T1 or T2, without reinjecting an input signal.
[0276] It will be noted that when the downmix T1 or T2 introduces a delay, the energy compensation must take this delay into account to align the energy of the downmix in function of the input channels. This can be achieved by shifting the input channels to synchronize them with the downmix. In variants, equivalently, we can also calculate the energy used for energy compensation by sub-blocks. For example, with a delay R = 1 ms, we can divide the 20 ms frame into 20 sub-blocks and take the energy of the last sub-block of the previous frame and the first 19 sub-blocks of the current frame for the input channels.
[0277] [Fig.6b] illustrates an alternative of the channel reduction processing method described with reference to Figure 6A.
[0278] The embodiment illustrated in Figure 6B differs from that of Figure 6A only by the fact that the detection of a transition is implemented according to a second embodiment. According to this second embodiment, such detection is implemented during the downmix step 610 T1 (or T2) instead of being applied separately to each of the input signals Xi(n) and x2(n), or on the current frame Xi(t,Z) and x2(t,Z). For this purpose, with reference to [Fig.5], the energy of each 2 ms sub-block between the input and output signal of blocks 502 and 503 is compared. If the energy has lowered a threshold, for example 3 dB, it is considered that a transient has been strongly attenuated, and in this case the downmix T3 is applied in the current frame. If the energy has not lowered a threshold, for example 3 dB, we consider that a transient has not been attenuated, and in this case we apply the downmix T1 (or T2) in the current frame.
[0279] Here again, when external information Fbinaurais explicitly indicates that the 2-channel input signal is binaural, i.e. Fbinaurais =1, it will be possible according to the invention to choose a particular downmix, for example the downmix T3.
[0280] [Fig.8] illustrates a channel reduction processing device 800, within the meaning of the invention.
[0281] The device 800 comprises a processing circuit typically including: - a memory MEM1 for storing instruction data of a computer program within the meaning of the invention; - an INT1 interface for receiving a stereo audio signal 1 x / n); U2(n) - a processor PROC1 for receiving this signal and processing it by executing the computer program instructions stored in the memory MEM1, with a view to carrying out channel reduction processing; in particular, the processor being capable of controlling the processing modules as described with reference to figures 2 to 5; and - a COM 1 communication interface for transmitting the reduced signals, resulting from the channel reduction processing, m(t, Z), to another processing module, for example a coding or decoding module of a signal coder or decoder audio.
[0282] Of course, this [Fig.8] illustrates an example of a structural embodiment of a channel reduction processing device within the meaning of the invention.
[0283] Figures 2 to 7 commented above describe in detail functional embodiments of this device.
[0284] Applications of this type of channel reduction processing are found, for example, in audio coding, for example when the remote party does not have the capacity to reproduce stereo sound on its terminal. In this case, it is not necessary to transport a stereo signal and this saves bandwidth. This type of process can be used during point-to-point conversations if one of the participants makes a stereo or binaural sound recording. This type of process can also be present during multi-party conferences: the conference bridge spatializes the scene by creating a stereo scene, but not all participants necessarily have stereo reproduction capabilities. It is then appropriate to perform a downmix of the scene in mono for these participants.
[0285] The invention can also be applied for audio decoding: in this case, it is the user's renderer (restitution module) which will perform the downmix of the stereo content to adapt to the user's reduced capacities (single loudspeaker for example).
Claims
Claims
1. Method for channel reduction processing of a stereophonic signal to obtain a monophonic signal comprising, for a current frame: - a first step of selecting a channel reduction processing between two methods using filtering of the stereophonic signal and rephasing of the signals, the selection being carried out according to a phase indicator, representative of a measurement of the degree of phase opposition between the frequency components of the channels of the stereo signal, - a second step of selection between one of the two methods, selected at the end of the first selection step and a channel reduction processing method without rephasing, the selection being carried out according to the value of an energy ratio between the channels of the stereo signal.
2. A method according to claim 1, wherein the energy ratio between the channels of the stereo signal is calculated per frequency band, the second selection step being carried out per frequency band.
3. The method of claim 2, wherein the channel reduction processing without rephasing is selected for the frequency bands concerned only in the case where the number of frequency bands concerned exceeds a threshold.
4. Method according to one of the preceding claims in which a third selection step is carried out between one of the two methods selected at the end of the second selection step and a third channel reduction processing method not using filtering of the stereophonic signal, said selection being implemented as a function of: - a presence or absence of at least one transition in the stereophonic signal, or - a quality level of the filtering of the stereophonic signal used during the implementation of processing using filtering of the stereophonic signal, or - an indication of the presence of a binaural signal.
5. A channel reduction processing method according to claim 4, wherein if a presence of said at least one transition is detected (613), respectively not detected (613), in the stereophonic signal, the channel reduction processing which does not use a filtering of the stereo signal, respectively which uses stereo signal filtering, is selected (612).
6. Channel reduction processing method according to claim 4, wherein the detection of said at least one transition is implemented before applying, to said stereophonic signal, one or the other of the two channel reduction processes.
7. Channel reduction processing method according to claim 1, wherein the detection of said at least one transition is implemented as follows: - calculating the energy of the stereophonic signal, called input energy, before applying the filtering, - calculating the energy of the monophonic signal, called output energy, after applying the filtering, - calculating a ratio between the input energy and the output energy, - comparing said ratio with an energy threshold, and - if said ratio is lower, respectively higher, than said energy threshold, the channel reduction processing which does not use filtering of the stereophonic signal, respectively which uses filtering of the stereophonic signal, is selected.
8. A channel reduction processing method according to claim 4, wherein said filtering quality level is compared with a filtering quality threshold, and if said filtering quality level is lower, respectively higher, than said filtering quality threshold, the channel reduction processing which does not use stereophonic signal filtering, respectively which uses stereophonic signal filtering, is selected.
9. Method according to one of the preceding claims, in which the method comprises a preliminary phase of adapting the channel reduction processing filters before application to the stereo signal by the following steps: - obtaining the impulse responses of the filters corresponding to the channel reduction processing; - truncation of a part of the impulse responses; - weighting of the remaining part by applying a pon- deration; - normalization of the impulse responses resulting from windowing to obtain the appropriate filters, to be applied to a stereo signal frame.
10. The method of claim 9, wherein the truncation comprises retaining a causal part and a non-causal part of the impulse responses.
11. A channel reduction processing device comprising a processing circuit for implementing the steps of the channel reduction processing method according to one of claims 1 to 10.
12. Storage medium, readable by a processor, storing a computer program comprising instructions for executing the method according to one of claims 1 to 10.
Citation Information
Patent Citations
Adaptive channel-reduction processing for encoding a multi-channel audio signal
WO2017103418A1
