Sensor array beamformer post-processor

• US 9,054,764 B2
• Filed: 07/20/2011
• Issued: 06/09/2015
• Est. Priority Date: 05/17/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Tashev, Ivan, Acero, Alejandro
• Primary Examiner(s): Saunders, Jr., Joseph
• Assistant Examiner(s): MOONEY, JAMES K
• Application Number:
  US13/187,235
• Publication Number:
  US 20110274289A1
• Time in Patent Office: 1,420 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: H04B 7/0854 using error minimizing algorithms, e.g. minimum mean squared error [MMSE], "cross-correlation" or matrix inversion

Abstract

A novel beamforming post-processor technique with enhanced noise suppression capability. The present beamforming post-processor technique is a non-linear post-processing technique for sensor arrays (e.g., microphone arrays) which improves the directivity and signal separation capabilities. The technique works in so-called instantaneous direction of arrival space, estimates the probability for sound coming from a given incident angle or look-up direction and applies a time-varying, gain based, spatio-temporal filter for suppressing sounds coming from directions other than the sound source direction, resulting in minimal artifacts and musical noise.

Claims

1. A computer-implemented process for improving the directivity and signal to noise ratio of the output of a beamformer employed with a sensor array in an environment, comprising:

capturing sound sources dispersed at locations in the environment with sensors of a sensor array;

inputting signals of the sound sources and a desired signal captured by the sensors of a sensor array in the frequency domain defined by frequency bins and frames in time;

computing a beamformer output as function of the input signals divided into frequency bins and frames in time;

dividing a spatial region corresponding to a working space of the sensor array into a plurality of incident angle regions, and for each frequency bin and incident angle region, computing the probability that the desired signal occurs at a given incident angle region using an instantaneous direction of arrival computation; and

spatially filtering the beamformer output by multiplying the probability that the desired signal occurs at a given incident angle region by the beamformer output while attenuating signals from the locations of the sound sources.

2. The computer-implemented process of claim 1 wherein the captured sound sources are estimated locations.

3. The computer-implemented process of claim 1 wherein the input signals in the frequency domain are converted from the time domain into the frequency domain prior to inputting them using a Modulated Complex Lapped Transform (MCLT).

4. The computer-implemented process of claim 1 wherein the sensors are microphones and wherein the sensor array is a microphone array.

5. The computer-implemented process of claim 1 wherein the instantaneous direction of arrival computation for each frequency bin is based on the phase differences of the input signals from a pair of sensors.

6. The computer-implemented process of claim 1 wherein spatially filtering the beamformer output attenuates signals originating from directions other than the direction of the desired signal.

7. A system for improving the signal to noise ratio of a desired signal received from a microphone array in an environment, comprising:

a general purpose computing device;

a computer program comprising program modules executable by the general purpose computing device, wherein the computing device is directed by the program modules of the computer program to,

capture audio signals of dispersed sound sources and a desired signal in an environment in the time domain with a microphone array;

convert the time-domain signals to frequency-domain and frequency bins using a converter;

input the signals in the frequency domain into a beamformer and compute a beamformer output wherein the beamformer output represents the optimal solution for capturing an audio signal at a target point using the total microphone array input;

estimate the probability that the desired signal comes from a given incident angle using an instantaneous direction of arrival computation; and

output an enhanced signal for the desired signal with a greater signal to noise ratio by taking the product of the beamformer output and the probability estimation that the desired signal comes from a given incident angle while attenuating audio signals that come from directions of the dispersed sound sources.

8. The system of claim 7 wherein the instantaneous direction of arrival computation for each frequency bin is based on the phase differences of the input signals from a pair of microphones.

9. The system of claim 7 wherein the beamformer is a time-invariant beamformer.

10. The system of claim 7 wherein the enhanced signal with a greater signal to noise ratio is computed and output in real time.

11. The system of claim 7 wherein the modules to estimate the probability that a desired signal comes from a given incident angle using an instantaneous direction of arrival computation and the module to output an enhanced signal with a greater signal to noise ratio by taking the product of the beamformer output and the probability estimation that the desired signal comes from a given incident angle form a post-processor that attenuates signals originating from directions other than the direction of the desired signal to output a signal with an enhanced signal to noise ratio.

12. A computer-implemented process for improving the signal to noise ratio of a desired signal received from a microphone array in an environment, comprising:

capturing audio signals of dispersed sound sources and a desired signal in the environment in the time domain with a microphone array;

converting the time-domain signals to frequency-domain and frequency bins using a converter;

inputting the signals in the frequency domain into a beamformer and computing a beamformer output wherein the beamformer output represents the optimal solution for capturing an audio signal at a target point using the total microphone array input;

estimating the probability that the desired signal comes from a given incident angle using an instantaneous direction of arrival computation; and

outputting an enhanced signal of the desired signal with a greater signal to noise ratio by taking the product of the beamformer output and the probability estimation that the desired signal comes from a given incident angle.

13. The computer-implemented process of claim 12 further comprising attenuating audio signals that come from directions of the sound sources.

14. The computer-implemented process of claim 12 wherein the instantaneous direction of arrival computation for each frequency bin is based on the phase differences of the input signals from a pair of microphones.

15. The computer-implemented process of claim 12 wherein the beamformer is a time-invariant beamformer.

16. The computer-implemented process of claim 12 wherein the enhanced signal with a greater signal to noise ratio is computed and output in real time.

17. The computer-implemented process of claim 12 wherein the instantaneous direction of arrival computation is based on the phase differences of the input signals from a pair of sensors.

18. The computer-implemented process of claim 12 wherein the dispersed sound sources are at estimated locations.