A decorrelator for decorrelating an input signal and a method thereof

Abstract

Decorrelating the input signal includes performing all-pass filtering to phase-shift a first input signal by a phase shift, and performing the all-pass filtering includes filtering using one or more subsequent controllable all-pass filter stages, each controllable all-pass filter stage having a filter quality and a cut-off frequency. Decorrelating also includes controlling at least one of the filter quality and the cut-off frequency of the controllable all-pass filter stage to vary over time.

Description

Technical Field

[0001] The present disclosure relates to systems and methods for decorrelating input signals (generally referred to as "systems"). Background Art

[0002] In some cases, such as in multi-channel adaptive systems, it may be desirable to use reference signals or input signals that are statistically independent of each other, i.e., have as high a decorrelation as possible between each other. For example, the continuously estimated room impulse response (RIR) of a channel adaptive system can be used to automatically identify and compensate for changes in the room for suppressing acoustic echo (AEC). When doing so, the RIR represented by the room transfer function between the speakers and microphones installed in the room is determined (e.g., calculated, estimated, etc.) and compared with the reference data previously determined in a reference room and stored. Then, the resulting spectral deviation forms the basis for determining the compensation filter, which subsequently makes it possible to ultimately create a subjectively always identical acoustic impression, regardless of the current acoustic conditions present in the room. As long as the multi-channel adaptive system only uses a single signal, e.g., only emits sound omnidirectionally, there is no problem in determining or using the adaptively estimated RIR. However, if the device operates in a stereo or generally in a multi-channel playback mode, in which, for example, many different signals that may be spatially vectorized are played, there may be an ambiguity between the adaptively determined RIRs, depending on the correlation between the signals used. In such a case, it will no longer be possible to use the method for automatically compensating for room changes mentioned at the beginning, which, as is known, relies on continuously determined RIRs.

[0003] Such ambiguities in RIR estimation can be prevented by ensuring that the various input signals to be played are sufficiently decorrelated from each other. Usually, the two channels of a stereo system are sufficiently decorrelated from each other, and thus in the case of pure stereo playback, such a problem does not occur. However, when using so-called "upmix" algorithms (such as, for example, Logic7 or Dolby Pro Logic), such a problem does occur. These algorithms generate multi-channel signals (e.g., a 5.1 signal from a stereo input signal), where the additional signals thus generated no longer have a high decorrelation from each other, and thus create a risk of ambiguity in RIR estimation. For this reason, a decorrelator must be employed. Therefore, there is a need for reliable decorrelation systems and methods. Summary of the Invention

[0004] An exemplary decorrelator for decorrelating an input signal includes a controllable all-pass filter arrangement configured to phase-shift a first input signal through phase shifting. The all-pass filter arrangement includes one or more controllable all-pass filter stages connected in series, and each controllable all-pass filter stage has a filter quality factor and a cut-off frequency. The decorrelator further includes a filter controller operatively connected to the controllable all-pass filter arrangement and configured to control at least one of the filter quality factor and the cut-off frequency of the controllable all-pass filter stages to vary over time.

[0005] An exemplary method for decorrelating an input signal includes performing all-pass filtering to phase-shift a first input signal through phase shifting. The performing all-pass filtering includes filtering using one or more successive controllable all-pass filter stages, each controllable all-pass filter stage having a filter quality factor and a cut-off frequency. The method further includes controlling at least one of the filter quality factor and the cut-off frequency of the controllable all-pass filter stages to vary over time.

[0006] After reviewing the following detailed description and the drawings (FIGs.), other systems, methods, features, and advantages will be apparent or will become apparent to those skilled in the art. It is intended that all such additional systems, methods, features, and advantages be included within this specification, be within the scope of the invention, and be protected by the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] The systems and methods may be better understood with reference to the following drawings and description. The components in the drawings are not necessarily to scale, but rather emphasis is placed upon illustrating the principles of the invention. In addition, like reference numerals designate corresponding parts throughout the different views.

[0008] Figure 1 is a schematic diagram showing an exemplary time-varying decorrelator, where the filter cut-off frequency is non-time-varying and the filter quality factor is time-varying.

[0009] Figure 2 is a schematic diagram showing a two-multiplier design of an M-th order all-pass filter.

[0010] Figure 3 is a Bode plot showing the magnitude and phase curves of two exemplary all-pass filter chains.

[0011] Figure 4 is a plot showing the group delay over frequency for each chain.

[0012] Figure 5 is a flowchart showing an example method for decorrelating an input signal.

[0013] Figure 6Signal flow diagram of an exemplary application of a decorrelator. Detailed implementation

[0014] Figure 1 An exemplary time-varying decorrelator is shown, where the filter cut-off frequency fc m ,(n) is non-time-varying, and the filter quality factor Q n (n) is time-varying, where n is the discrete-time parameter m = [1,…,M] and M = the (integer) number of all-pass filter stages included in the decorrelator. For example, M second-order all-pass filter stages AP2 can be connected in series to form a chain 101 of all-pass filter stages AP2, where the filter controller 102 controls the filter quality factor Q n (n) of each all-pass filter stage AP2 to vary with time. Alternatively, the quality factor Q m (n) is non-time-varying, and the cut-off frequency fc m (n) is time-varying. In this case, the poles (whose spectral positions in the unit circle are determined only by the fundamental frequency of the filter) can thus be distributed, for example, non-linearly at frequencies similar to those of the human ear, which makes sense from a psychoacoustic perspective. The decorrelator receives the input signal x(n) to be decorrelated and provides the decorrelated signal y(n).

[0015] Additionally or alternatively, the fundamental frequency of the filter with a maximum frequency of fs / 4 should be selected only to ensure that the resulting group delay of the all-pass filter chain not only rises to this frequency only due to the accumulation of the continuously decreasing individual phase responses, but also starts to decrease again after reaching the maximum frequency fs / 4, thus avoiding excessive and unwanted accumulation of the group delay. Whether or not this is the case, the above options can be used, as well as the option where two filter parameters (i.e., the cut-off frequency fc n (n) and the quality factor Q n (n)) are time-varying.

[0016] For example, a lattice ladder filter provides a simple way to implement an M-order parametric all-pass filter stage. There are various designs of lattice ladder filters, such as, for example, one-multiplier, two-multiplier, and four-multiplier designs. In an all-pass filter, the attenuation of the filter is constant at all frequencies, but the relative phase between the input and the output varies with frequency. Figure 2 An exemplary signal flow of a two-multiplier design of an M-order all-pass filter is shown. As can be seen from Figure 2It can be seen that an exemplary all-pass filter stage with a lattice design includes a plurality of lattice stages 201, 202, and 203, and each of the lattice stages has the same basic structure. Each single stage 201, 202, 203 has a forward path input, a forward path output, a backward path input, and a backward path output. The forward path input is operatively coupled to one input of forward adders 204, 205, 206, and the output of the adder serves as the forward path output. The backward path input is operatively coupled to one input of backward adders 210, 211, 212 via time delays 207, 208, 209, and the output of the backward adder serves as the backward path output. The other input of forward adders 204, 205, 206 is operatively coupled to the backward path input via first multipliers 213, 214, 215 and time delays 207, 208, 209. The other input of backward adders 210, 211, 212 is operatively coupled to the forward path output via second multipliers 216, 217, 218.

[0017] The forward path input of stage 201 receives the filter input signal x(n) = f M (n), and provides the filter output signal x'(n) = g M (n) at its backward path output. In addition, the backward path input of stage 201 receives the signal g M-1 (n), and provides the signal f M-1 (n) at its forward path output. For example, if n = 3, the signal g M-1 (n) is g2(n), and the signal f N-1 (n) is f2(n). In Figure 2 the example shown, the signal g2(n) is provided at the backward path output of lattice stage 202, and the signal f2(n) is received at the forward path input of lattice stage 202. In addition, lattice stage 201 provides the signal f2(n) at its forward path output, and this signal is sent to the forward path input of lattice stage 203; and it receives the signal g1(n) from the backward path output of lattice stage 203 at its backward path input. The forward path output of lattice stage 203 provides the signal f0(n), and this signal serves as the signal g0(n) supplied to the backward path input of lattice stage 203.

[0018] The advantage of lattice filters is that their filter coefficients correspond to reflection coefficients, and the reflection coefficients can be determined, for example, using the Levinson Durbin Recursion. One of the properties of the reflection coefficients is that the reflection coefficients ensure that the filter is stable as long as the value of the reflection coefficient remains less than 1, that is, as long as K m ≤ │1│, where m = 1,..., M, and M is the order of the filter.

[0019] In the case of a second-order lattice all-pass filter, the first filter (or reflection) coefficient K1 corresponds to the filter cut-off frequency fc, and the second filter coefficient K2 corresponds to the filter quality factor Q. Taking advantage of this, the filter coefficient K c can be easily generated over time, for example, by a common pseudo-random number generator (white noise generator) that provides quasi-random values from the range [–1,…,+1]. According to the following, the range of values used can be further restricted, for example, in order to prevent the filter quality factor from becoming too large:

[0020] K2(n) 1，...M ∈[0,...,K2 Max ,

[0021] where K2 Max ≤1 and M is the number of all-pass filters in the chain.

[0022] In order to prevent the generation of disturbing acoustic artifacts, the dynamics over time of one or more time-varying filter parameters or one or more filter coefficients are restricted, i.e., one or more time-varying filter parameters or one or more filter coefficients do not change too much. To achieve this, the dynamic range in which the one or more filter parameters (fc and / or Q) under discussion can change from one sample to the next is restricted accordingly (e.g., fc can change by no more than Δfc = 1 [Hz] from one sample to the next), or the duration for which one or more filter parameters can change infinitely is very long, in which case interpolation can be performed in between.

[0023] Here, the advantage of using a lattice filter to implement an all-pass filter and the accompanying reflection filter coefficients becomes apparent again, because using this structure allows for direct parameter changes in the filter coefficients. In contrast, when using a common all-pass filter, such as a common all-pass filter using a direct form structure, the filter coefficients must be continuously recalculated from finite or interpolated filter parameters, which requires a considerable amount of computational effort, while using a lattice filter does not require said considerable amount of computational effort.

[0024] In practice, an update time of approximately t ud = 1 [s] can be useful. For example, at each t ud , new time-varying filter coefficients K2 c (where c = 1,..., C, and C is the number of second-order all-pass filters) are calculated by a pseudo-random number generator from a series of K2 c ∈[0,…,K2 max and applied. Then at t udInterpolate these filter coefficients (e.g., linearly) over a determined time period so that at t ud end, all time-varying filter coefficients K2 c (n) correspond to new values generated by a pseudo-random number generator. In this simple way and without unduly increasing the computational effort, interference acoustic artifacts can be greatly reduced so that they no longer pose an acoustic problem.

[0025] Figure 3 is a Bode plot showing the magnitude curves ( s upper curve in Figure 3 ) and phase curves ( Figure 3 lower curve in s ) of two exemplary all-pass filter chains operating at a sampling rate f of 16 [kHz] and each chain comprising 16 second-order all-pass filter stages. The filter cut-off frequencies are limited to a frequency band between 100 [Hz] and f s / / 2 - f max 8 [Hz] and can be linear or distributed within this range according to a psychoacoustic scale (e.g., Bark scale). The maximum allowable quality factor is determined by K2 c = 0.99 and the time-varying filter parameter K2 max (n) ∈ [0, …, K2 Figure 3 . The interpolation of the time-varying filter parameter is performed linearly, and the signals to be decorrelated are the left and right channel signals of a multi-channel signal, where the central channel signal is not processed. The left channel signal is fed to one all-pass filter chain, and the right channel is fed to another all-pass filter chain. As can be seen from the

[0026] Figure 4 upper curve of s , the level degradation caused by the all-pass filter chain is negligible.

[0027] Referring to Figure 5 , an exemplary decorrelation method for decorrelating an input signal includes performing all-pass filtering to phase-shift a first input signal x(n) by a phase shift, said performing all-pass filtering including filtering with one or more successive controllable all-pass filter stages, each controllable all-pass filter stage having a filter quality and a cut-off frequency (procedure 501). The method further includes controlling at least one of the filter quality (procedure 502) and the cut-off frequency (procedure 503) of the controllable all-pass filter stage to vary with time.

[0028] Figure 6is a signal flow diagram of an exemplary application of a decorrelator. As Figure 6 shown, an upmixer 601 using an upmixing algorithm can extract a center signal C(n) from two stereo input signals L(n) and R(n). Then, these three signals are decorrelated in a decorrelator 602 and are directed in various directions in a room using corresponding beamforming filters of a beamformer 603, where the extracted center signal C(n) is directed towards a listening position, and the two stereo signals L(n) and R(n) are transmitted in opposite directions, i.e., backwards, ideally to the position where a solid wall is located, thereby creating a specific acoustic effect from the resulting diffusion. In one option, the extracted center signal C(n) is decorrelated because the two stereo signals L(n) and R(n) may already be sufficiently uncorrelated with respect to each other and can thus be used as such for beamforming. Alternatively, since decorrelation can further increase the diffusion of these signals, instead of decorrelating the direct sound (generated from the two stereo signals), i.e., the center channel, the two effect channels, i.e., the two stereo signals L(n) and R(n), are decorrelated.

[0029] In another example, the all-pass filter parameters, cut-off frequency, and / or quality factor are controlled depending on a correlation analysis of an input signal and at least one comparison signal (e.g., other input or reference signal) so that decorrelation is applied only when a certain correlation between the reference signals is detected (e.g., in certain spectral ranges). The filter controller 102 shown in the figure can be adapted to execute this procedure. For example, a processor implementing the filter controller 102 includes software that allows evaluating a value corresponding to the degree of correlation and comparing this value with a threshold.

[0030] In some applications, such as in multi-channel adaptive systems (such as multi-channel acoustic echo cancellers (MCAECs)), it may be advantageous to decorrelate reference signals so that these reference signals become statistically independent and thus allow for different (i.e., distinct) estimates of the "actual" room impulse response (RIR). For example, this applies to automatic equalization systems that are designed to compensate for different room characteristics in order to ideally achieve a subjectively similar tonal balance, regardless of the room in which the device is used and / or the position of the device in the room.

[0031] If a single signal is used as a reference, the above disadvantages do not exist. If a stereo signal is used as a reference, there are generally no negative effects either, since a typical stereo input signal provides a sufficiently high degree of decorrelation between its left and right channels. However, if an upmixing algorithm is used to create several signals based on a (mainly) stereo input, we do encounter the problem of ambiguity if no further measures are taken to decorrelate its output signals, which may be used as reference signals for an MCAEC. In such a case, it is necessary to introduce additional decorrelation to one or more output signals of the upmixer before they are used as references for an MCAEC.

[0032] The systems and methods described above provide a simple and effective way of implementing a decorrelator which moreover does not produce significant additional acoustic artifacts. An all-pass filter (AP) chain is used, including for example parametric filters, to achieve a simple time-variation of certain parameters such as its filter quality and / or its cut-off frequency. In addition, a set of fixed cut-off frequencies distributed over a certain limited frequency range can be used in combination with a time-varying quality factor, which is also restricted to a defined adjustable range, to avoid acoustic artifacts which may occur if too high a quality factor value is employed.

[0033] The methods described above can be encoded in a computer-readable medium as instructions for a processor to execute, the computer-readable medium such as a CD ROM, disk, flash memory, RAM or ROM, electromagnetic signal or other machine-readable medium. Alternatively or additionally, any type of logic can be utilized and any type of logic can be implemented as analog or digital logic using hardware such as one or more integrated circuits including amplifiers, adders, delayers and filters or one or more processors which execute instructions for amplification, addition, delay and filtering; or any type of logic can be implemented in the form of software in an application programming interface (API) or a dynamic link library (DLL) as a function available in shared memory or defined as a local or remote procedure call; or any type of logic can be implemented as a combination of hardware and software.

[0034] The method may be implemented by software and / or firmware stored on or in a computer-readable medium, a machine-readable medium, a propagated signal medium, and / or a signal-bearing medium. The medium may include any device that contains, stores, communicates, propagates, or transports executable instructions for use by or in connection with an instruction-executable system, apparatus, or device. The machine-readable medium may optionally be, but is not limited to, an electronic signal, a magnetic signal, an optical signal, an electromagnetic signal, or an infrared signal, or a semiconductor system, apparatus, device, or propagation medium. A non-exhaustive list of examples of machine-readable media includes: magnetic or optical disks, volatile memory such as random access memory "RAM", read-only memory "ROM", erasable programmable read-only memory (i.e., EPROM), or flash memory, or optical fiber. The machine-readable medium may also include a tangible medium on which executable instructions are printed, since the logic may be stored electronically as an image or in another format (e.g., by optical scanning), and then compiled and / or interpreted or otherwise processed. The processed medium may then be stored in a computer and / or machine memory.

[0035] The system may include additional or different logic and may be implemented in many different ways, including a controller that implements a filter chain and / or a filter controller. The controller may be implemented as a microprocessor, a microcontroller, an application-specific integrated circuit (ASIC), discrete logic, or a combination of other types of circuits or logic. Similarly, the memory may be DRAM, SRAM, flash memory, or other types of memory. Parameters (e.g., conditions and thresholds) and other data structures may be stored and managed separately, incorporated into a single memory or database, or organized logically and physically in many different ways. Programs and instruction sets may be part of a single program, separate programs, or distributed across several memories and processors.

[0036] A description of embodiments has been presented for purposes of illustration and description. Appropriate modifications and changes to the embodiments may be made in light of the above description, or may be obtained by practicing the methods. For example, unless otherwise indicated, one or more of the described methods may be performed by a suitable device and / or combination of devices. The described methods and associated actions may also be performed in various orders other than the order described in this application, in parallel, and / or simultaneously. The described system is exemplary in nature and may include additional elements and / or omit elements.

[0037] As used in this application, an element or step recited in the singular and preceded by the word "a" or "an" should be understood to not exclude a plurality of such elements or steps, unless such exclusion is specified. Additionally, a reference to "one embodiment" or "one example" of the present disclosure is not intended to be construed as excluding the existence of additional embodiments that also incorporate the recited features. The terms "first," "second," and "third," etc. are used only as labels and are not intended to impose numerical requirements or a particular positional order on their objects.

[0038] Although various embodiments of the invention have been described, those of ordinary skill in the art will appreciate that many embodiments and implementations are possible within the scope of the invention. Specifically, those skilled in the art will recognize the interchangeability of various features from different embodiments. Although these techniques and systems have been disclosed in the context of certain embodiments and examples, it will be understood that these techniques and systems can be extended beyond the specifically disclosed embodiments to other embodiments and / or uses and their obvious modifications.

Claims

1. A decorrelator for decorrelating an input signal, the decorrelator comprising: A controllable all-pass filter arrangement configured to phase-shift the input signal by a phase shift, the all-pass filter arrangement comprising one or more controllable all-pass filter stages connected in series, and each controllable all-pass filter stage having a filter quality factor and a cut-off frequency; And A filter controller operably connected to the controllable all-pass filter arrangement and configured to control at least one of the filter quality factor and the cut-off frequency of the controllable all-pass filter stage to vary over time, Wherein the input signal phase-shifted by the controllable all-pass filter corresponds to at least one output of an upmixer; Wherein the filter quality factor is restricted within a given range; Wherein the given range of the filter quality factor is adjustable, and Wherein the input signal phase-shifted by the controllable all-pass filter includes the at least one reference signal through the multi-channel acoustic echo canceller.

2. The decorrelator according to claim 1, wherein the cut-off frequency is fixed and the filter quality factor is time-varying.

3. The decorrelator according to claim 2, wherein the cut-off frequencies are distributed within a restricted frequency range.

4. The decorrelator according to claim 3, wherein the cut-off frequencies are distributed according to a psychoacoustic scale.

5. The decorrelator according to claim 1, wherein the all-pass filter stage has a parametric filter structure.

6. The decorrelator according to claim 1, wherein the all-pass filter stage has a lattice-ladder filter structure.

7. The decorrelator according to claim 1, wherein the filter controller includes a random generator configured to generate a random control signal to control at least one of the filter quality factor and the cut-off frequency of the controllable all-pass filter stage.

8. The decorrelator according to claim 1, wherein the filter controller is configured to detect the correlation between the input signal and at least one comparison signal and control at least one of the filter quality factor and the cut-off frequency of the controllable all-pass filter stage depending on the detected correlation.

9. The decorrelator according to claim 1, wherein at least one of the filter quality and the cut-off frequency is interpolated over time.

10. A decorrelating method for decorrelating an input signal, the decorrelating method comprising: Performing all-pass filtering to phase-shift the input signal by a phase shift, the performing all-pass filtering including filtering with one or more successive controllable all-pass filter stages, each controllable all-pass filter stage having a filter quality factor and a cut-off frequency; And Controlling at least one of the filter quality factor and the cut-off frequency of the controllable all-pass filter stage to vary over time, Wherein the input signal phase-shifted by the all-pass filtering corresponds to at least one output of an upmixer; wherein, the filter quality factor is limited within a given range; wherein, the given range of the filter quality factor is adjustable; and wherein, the input signal that undergoes phase shift through the all-pass filtering includes the at least one reference signal for the multi-channel acoustic echo canceller.

11. The decorrelation method according to claim 10, wherein the cut-off frequency is distributed within a restricted frequency range.

12. The decorrelation method according to claim 10, wherein the all-pass filter stage has a parametric filter structure.

13. The decorrelation method according to claim 10, wherein the all-pass filter stage has a lattice-ladder filter structure.

14. The decorrelation method according to claim 10, wherein controlling the all-pass filter stage includes generating a random control signal for controlling at least one of the filter quality factor and the cut-off frequency of the controllable all-pass filter stage.

15. The decorrelation method according to claim 10, wherein controlling the all-pass filter stage includes detecting the correlation between the input signal and at least one comparison signal; and controlling at least one of the filter quality factor and the cut-off frequency of the controllable all-pass filter stage depending on the detected correlation.

16. The decorrelation method according to claim 10, wherein at least one of the filter quality and the cut-off frequency is interpolated over time.

17. A computer program product, the computer program product comprising instructions that, when executed by a computer, cause the computer to perform the steps of the method according to any one of claims 10-16.

Images