Signal quality beam selection

Patent metadata

Field	Value
Publication number	US-9973849-B1
Application number	US-201715710382-A
Country	US
Kind code	B1
Filing date	Sep 20, 2017
Priority date	Sep 20, 2017
Publication date	May 15, 2018
Grant date	May 15, 2018

Abstract

A device that includes a first beam selection component that selects a portion of the multiple beams prior to performing adaptive noise cancellation, in addition to a second beam selection component that selects a beam after adaptive noise cancellation is performed. The device only performs adaptive noise cancellation on the portion of the multiple beams that are selected, reducing a complexity of performing adaptive noise cancellation. In addition, as the first beam selection component selects from the multiple beams before adaptive noise cancellation, the quality of the signals are not reduced and/or desired signals are not suppressed, resulting in the first beam selection component increasing a likelihood that the second beam selection component selects the desired beam. The first beam selection component may use a different algorithm and/or selection criteria than the second beam selection component in order to select beams associated with desired signals such as speech.

First claim

What is claimed is:

1. A computer-implemented method comprising:
receiving input audio data, the input audio data corresponding to input audio captured by a microphone array;
performing beamforming on the input audio data to determine a plurality of beamformed audio data, the plurality of beamformed audio data including:
first beamformed audio data corresponding to a first direction,
second beamformed audio data corresponding to a second direction, and
third beamformed audio data corresponding to a third direction;

determining a first signal-to-interference ratio (SIR) associated with the first beamformed audio data;
determining a second SIR associated with the second beamformed audio data;
determining a third SIR associated with the third beamformed audio data;
determining that the first SIR is above a threshold value that corresponds to a strong audio signal;
determining that the second SIR is above the threshold value;
determining that the third SIR is below the threshold value;
performing adaptive noise cancellation on the first beamformed audio data to generate first modified beamformed audio data;
performing adaptive noise cancellation on the second beamformed audio data to generate second modified beamformed audio data;
determining a first signal-to-noise ratio (SNR) associated with the first modified beamformed audio data;
determining a second SNR associated with the second modified beamformed audio data;
determining that the first SNR is greater than the second SNR; and
sending the first modified beamformed audio data for further processing.

2. The computer-implemented method of claim 1 , further comprising:
selecting a first portion of the plurality of beamformed audio data as a first noise reference signal;
selecting a second portion of the plurality of beamformed audio data as a second noise reference signal;
combining the first noise reference signal and the second noise reference signal to generate interference data; and
determining the first SIR by dividing a first power value associated with the first beamformed audio data by a second power value associated with the interference data.

3. The computer-implemented method of claim 1 , further comprising:
determining a first difference between the first SIR and the second SIR;
determining a second difference between the first SIR and the third SIR;
determining that the first difference is below a second threshold value, the second threshold value representing that SIR values are similar in magnitude;
determining that the second difference is above the second threshold value; and
performing the adaptive noise cancellation on the first beamformed audio data and the second beamformed audio data.

4. The computer-implemented method of claim 1 , further comprising, prior to performing adaptive noise cancellation on the first beamformed audio data:
determining a fourth SIR associated with fourth beamformed audio data of the plurality of beamformed audio data;
determining that the fourth SIR is above the threshold value;
selecting a first noise reference signal from the plurality of beamformed audio data;
determining a first correlation value representing a first correlation indicating a similarity between the first beamformed audio data and the first noise reference signal;
selecting second noise reference signal from the plurality of audio data;
determining a second correlation value representing a second correlation between the fourth beamformed audio data and the second noise reference signal;
determining that the first correlation value is below a second threshold value, the second threshold value representing whether signals are similar;
determining that the second correlation value is above the second threshold value;
performing adaptive noise cancellation on the first beamformed audio data in response to the first correlation value being below the second threshold value; and
determining not to perform adaptive noise cancellation on the fourth beamformed audio data in response to the second correlation value being above the second threshold value.

5. A computer-implemented method comprising:
receiving input audio data corresponding to input audio captured by a microphone array;
determining, from the input audio data, a plurality of audio signals, each audio signal of the plurality of audio signals corresponding to a respective direction;
selecting, from the plurality of audio signals, a first audio signal corresponding to a first direction and a second audio signal corresponding to a second direction, wherein the selecting uses at least one signal quality metric;
performing adaptive noise cancellation on the first audio signal to generated a modified first audio signal;
performing adaptive noise cancellation on the second audio signal to generate a modified second audio signal;
determining a second signal quality metric associated with the modified first audio signal;
determining a third signal quality metric associated with the modified second audio signal;
determining that the second signal quality metric indicates a better signal quality than the third signal quality metric; and
outputting the modified first audio signal.

6. The computer-implemented method of claim 5 , further comprising:
determining a first number of audio signals of the plurality of audio signals, each of the first number of audio signals having a higher signal quality metric than each of a remaining second number of audio signals of the plurality of audio signals;
and
performing adaptive noise cancellation on the first number of audio signals, the first number of audio signals including the first audio signal and the second audio signal.

7. The computer-implemented method of claim 5 , further comprising:
determining a fourth signal quality metric associated with the first audio signal;
determining a fifth signal quality metric associated with the second audio signal;
determining a sixth signal quality metric associated with a third audio signal of the plurality of audio signals, the third audio signal corresponding to a third direction;
selecting, based on the fourth signal quality metric, the first audio signal for further processing;
selecting, based on the fifth signal quality metric, the second audio signal for further processing; and
determining, based on the sixth signal quality metric, not to select the third audio signal for further processing.

8. The computer-implemented method of claim 7 , further comprising:
determining a first difference between the first signal quality metric and the second signal quality metric;
determining a second difference between the first signal quality metric and the third signal quality metric;
determining that the first difference is below a threshold value;
determining that the second difference is above the threshold value; and
determining, based on the first difference and the second difference, a first number of the plurality of audio signals to select for further processing.

9. The computer-implemented method of claim 5 , further comprising:
selecting a first noise reference signal from the plurality of audio signals;
selecting a second noise reference signal from the plurality of audio signals;
determining, based on the first noise reference signal and the second noise reference signal, an interference signal; and
determining a fourth signal quality metric by dividing a first power value associated with the first audio signal by a second power value associated with the interference signal.

10. The computer-implemented method of claim 5 , further comprising:
determining a fourth signal quality metric associated with the first audio signal

Assignees

• Amazon Tech Inc

Inventors

• Zhang Xianxian
• Chhetri Amit Singh
• Ovaert Laurent Andre

Classifications

• H04R2410/01

  Noise reduction using microphones having different directional characteristics · CPC title

• H04M3/2236

  Quality of speech transmission monitoring · CPC title

• H04R29/005

  Microphone arrays · CPC title

• H04R3/005 Primary

  for combining the signals of two or more microphones (specially adapted for hearing aids H04R25/407) · CPC title

• H04R2430/25

  Array processing for suppression of unwanted side-lobes in directivity characteristics, e.g. a blocking matrix · CPC title