Auditory source separation in a spiking neural network

1. A method of audio source segregation in an artificial neural network, comprising:

selecting an audio attribute of an audio signal;

representing a portion of the audio attribute dominated by a single source as at least one source spiking event in the artificial neural network, in which the single source comprises an anchor that provides an indication of a channel of the audio signal, the anchor processed using at least one band pass filter to identify a band pass filter output within a specified range;

representing a remaining portion of the audio signal as at least one audio signal spiking event in the artificial neural network; and

determining whether the remaining portion coincides with the single source based at least in part on coincidence of timing of the at least one source spiking event and the at least one audio signal spiking event.

2. The method of claim 1 , in which the determining is based at least in part on the at least one source spiking event and the at least one audio signal spiking event occurring within a temporal window.

3. The method of claim 1 , in which the audio attribute is pitch.

4. The method of claim 1 , in which the audio attribute is spatial location.

5. The method of claim 1 , in which the determining is based at least in part on a rate of speaking.

6. The method of claim 1 , in which the processing is conducted on a periodic basis.

7. An apparatus for audio source segregation in an artificial neural network, comprising:

a memory; and

at least one processor coupled to the memory, the at least one processor being configured:

to select an audio attribute of an audio signal;

to represent a portion of the audio attribute that is dominated by a single source as at least one source spiking event in the artificial neural network, in which the single source comprises an anchor that provides an indication of a channel of the audio signal, the anchor processed using at least one band pass filter to identify a band pass filter output within a specified range;

to represent a remaining portion of the audio signal as at least one audio signal spiking event in the artificial neural network; and

to determine whether the remaining portion coincides with the single source based at least in part on coincidence of timing of the at least one source spiking event and the at least one audio signal spiking event.

8. The apparatus of claim 7 , in which the at least one processor is further configured to determine whether the remaining portion coincides with the single source based at least in part on the at least one source spiking event and the at least one audio signal spiking event occurring within a temporal window.

9. The apparatus of claim 7 , in which the audio attribute is pitch.

10. The apparatus of claim 7 , in which the audio attribute is spatial location.

11. The apparatus of claim 7 , in which the at least one processor is further configured to determine whether the remaining portion coincides with the single source based at least in part on a rate of speaking.

12. The apparatus of claim 7 , in which the at least one processor is further configured to process the anchor on a periodic basis.

13. An apparatus for audio source segregation in an artificial neural network, comprising:

means for selecting an audio attribute of an audio signal;

means for representing a portion of the audio attribute that is dominated by a single source as at least one source spiking event in the artificial neural network, in which the single source comprises an anchor that provides an indication of a channel of the audio signal; the anchor processed using at least one band pass filter to identify a band pass filter output within a specified range;

means for representing a remaining portion of the audio signal as at least one audio signal spiking event in the artificial neural network; and

means for determining whether the remaining portion coincides with the single source based at least in part on coincidence of timing of the at least one source spiking event and the at least one audio signal spiking event.

14. The apparatus of claim 13 , in which the determining means further determines whether the remaining portion coincides with the single source based at least in part on the at least one source spiking event and the at least one audio signal spiking event occurring within a temporal window.

15. The apparatus of claim 13 , in which the audio attribute is pitch.

16. A non-transitory computer readable medium having encoded thereon program code for audio source segregation in an artificial neural network, the program code being executed by a processor and comprising:

program code to select an audio attribute of an audio signal;

program code to represent a portion of the audio attribute that is dominated by a single source as at least one source spiking event in the artificial neural network, in which the single source comprises an anchor that provides an indication of a channel of the audio signal, the anchor processed using at least one band pass filter to identify a band pass filter output within a specified range;

program code to represent a remaining portion of the audio signal as at least one audio signal spiking event in the artificial neural network; and

program code to determine whether the remaining portion coincides with the single source based at least in part on coincidence of timing of the at least one source spiking event and the at least one audio signal spiking event.

17. The non-transitory computer readable medium of claim 16 , further comprising program code to determine whether the remaining portion coincides with the single source based at least in part on the at least one source spiking event and the at least one audio signal spiking event occurring within a temporal window.

18. The non-transitory computer readable medium of claim 16 , in which the audio attribute is pitch.