Spatial decorrelation of the frequency domain of audio signals using a compatible method.

Abstract

The audio device includes a receiver 101 configured to receive a set of input audio signals segmented by a segmenter 103. For each segment, an output signal generator 105 generates an output audio signal by performing the following steps: generating a frequency bin representation of the input audio signal (201); generating a frequency bin representation of the output audio signal as a weighted combination of the frequency bin values ​​of the input audio signal for each frequency bin (203); and converting the frequency bin representation to the time domain. An adapter 107 updates the weights for a given weight according to a correlation measure between the output frequency bin values ​​for the generated output audio signal and the weighted input signal linked to the output audio signal. This approach typically provides improved frequency-domain adapted spatial decorrelation of the audio signal.

Claims

1. A speech decorrelation device for generating a set of output speech signals which are decorrelated signals of a set of input speech signals, wherein the speech decorrelation device is A receiver that receives the set of input audio signals, A segmenter that segments the aforementioned set of input audio signals into time segments, An output signal generator that generates the set of output audio signals, wherein each output audio signal in the set of output audio signals is linked to an input audio signal in the set of input audio signals, and the output signal generator generates for each time segment, A step of generating a frequency bin representation of the set of input audio signals, wherein each frequency bin of the frequency bin representation of the set of input audio signals includes a frequency bin value for each of the input audio signals in the set of input audio signals. A step of generating a frequency bin representation of the set of output audio signals, wherein each frequency bin of the frequency bin representation of the set of output audio signals includes a frequency bin value for each of the set of output audio signals, and the frequency bin value for a given output audio signal of the set of output audio signals for a given frequency bin is generated as a weighted combination of the frequency bin values ​​of the set of input audio signals for a given frequency bin. A step of generating a time-domain representation for each output audio signal from the frequency bin representation of the set of output audio signals. An output signal generator that performs the following: An adapter that updates the weights of the aforementioned weighted combinations Equipped with, The adapter updates a first weight for the contribution of the first frequency bin of the first frequency bin to the first input audio signal, linked to the first input audio signal, from the second frequency bin of the first frequency bin for the second input audio signal linked to the second input audio signal, depending on a correlation scale between the first previous frequency bin value of the first output audio signal for the first frequency bin and the second previous frequency bin value of the second output audio signal for the first frequency bin for the first frequency bin, A speech decorrelator wherein the adapter further updates a second weight which is the contribution from a third frequency bin value to the first frequency bin value, which is the frequency bin value of the first frequency bin for a first input audio signal, depending on the magnitude of the first previous frequency bin value.

2. The speech decorrelator according to claim 1, wherein the adapter updates the first weight according to the product of a first value and a second value, the first value being one of the first previous frequency bin value and the second previous frequency bin value, and the second value being the complex conjugate of the other of the first previous frequency bin value and the second previous frequency bin value.

3. The audio decorrelation device according to claim 1 or 2, wherein the adapter restricts the weight of the contribution from a third frequency bin value, which is the frequency bin value of the first frequency bin for the first input audio signal, to the first frequency bin value to be a real value.

4. The audio decorrelator according to any one of claims 1 to 3, wherein the adapter is configured such that a third weight, which is the weight for the contribution of the first frequency bin to the fourth frequency bin value of the second output audio signal from the first input audio signal, is the complex conjugate of the first weight.

5. The speech decorrelator according to any one of claims 1 to 4, wherein the weights of the weighted combinations for input audio signals other than the first input audio signal are complex numerical weights.

6. The adapter outputs an output bin value for the given frequency bin ω. y(ω)=W(ω)x(ω) The audio decorrelator according to any one of claims 1 to 5, wherein y(ω) is a vector containing the frequency bin values ​​for the output audio signal for the given frequency bin ω, x(ω) is a vector containing the frequency bin values ​​for the input audio signal for the given frequency bin ω, and W(ω) is a matrix having rows containing the weights of the weighted combination for the output audio signal.

7. The adapter is the weight w of the matrix W(ω) ｉｊ of, w ｉｊ (k+1,ω)=w ｉｊ (k,ω)-η(k,ω)[] ｉ (k, ω)y ｊ ＊ (k, ω)] The speech decorrelator according to any one of claims 1 to 6, wherein the device is fitted according to the formula, where i is the row index of the matrix W(ω), j is the column index of the matrix W(ω), k is the time segment index, ω represents the frequency bin, and η(k,ω) is a scaling parameter for fitting the fitting speed.

8. The audio decorrelator according to any one of claims 1 to 7, wherein the adapter compensates for the correlation value with respect to the signal level of the first frequency bin.

9. The speech decorrelator according to any one of claims 1 to 8, wherein the adapter is configured to initialize the weights for the weighted combination such that it includes at least one zero weight and one non-zero weight.

10. The speech decorrelator according to any one of claims 1 to 9, wherein the weighted combination is applied to the frequency representation of the weights formed by the weights for the first input speech signal and the second input speech signal for different frequency bins.

11. A voice device comprising the voice decorrelation device according to any one of claims 1 to 10, wherein the voice device is: A sound device further comprising a sound beamformer that receives a set of output sound signals and performs sound beamforming to generate a beamformed sound output signal.

12. The audio device according to claim 11, wherein the audio beamforming is adaptive audio beamforming.

13. A method for generating a set of output audio signals which are the uncorrelated signals of a set of input audio signals, wherein the method is: The steps include receiving the set of input audio signals, The steps include segmenting the set of input audio signals into time segments, A step of generating the set of output audio signals, wherein each output audio signal in the set of output audio signals is linked to one input audio signal in the set of input audio signals, and the step of generating the set of output audio signals is performed for each time segment. A step of generating a frequency bin representation of the set of input audio signals, wherein each frequency bin of the frequency bin representation of the set of input audio signals includes a frequency bin value for each of the input audio signals in the set of input audio signals. A step of generating a frequency bin representation of the set of output audio signals, wherein each frequency bin of the frequency bin representation of the set of output audio signals includes a frequency bin value for each of the output audio signals, and the frequency bin values ​​of the set of output audio signals for a given frequency bin are generated as a weighted combination of the frequency bin values ​​of the set of input audio signals for a given frequency bin. A step of generating a time-domain representation for each output audio signal from the frequency bin representation of the set of output audio signals. The step of generating the set of output audio signals includes performing the following: The method further includes updating the weights of the weighted combination, which includes updating a first weight for the contribution of the first frequency bin of the first frequency bin for the first output audio signal linked to the first input audio signal to the first input audio signal, from the second frequency bin of the first frequency bin for the second input audio signal linked to the second output audio signal, according to a correlation scale between a first previous frequency bin value of the first output audio signal for the first frequency bin and a second previous frequency bin value of the second output audio signal for the first frequency bin, A method wherein the step of updating the weights of the weighted combination includes updating a second weight which is the contribution from a third frequency bin value to the first frequency bin value, which is the frequency bin value of the first frequency bin for the first input audio signal, in proportion to the magnitude of the first previous frequency bin value.

14. A computer program comprising computer program code means, wherein when the computer program is executed on a computer, the computer program code means performs all the steps of the method according to claim 13.

Images