Keyword-based audio source localization

1. An apparatus comprising:

one or more processors; and

memory storing instructions that, when executed by the one or more processors, cause the apparatus to:

receive, from a plurality of microphones, one or more indications of first speech detected in a plurality of listening zones;

determine, based on the one or more indications of the first speech, that the first speech comprises a keyword;

determine a direction associated with the keyword;

detect, using an acoustic beam formed by at least some of the plurality of microphones, pointed in the direction associated with the keyword, and having a width that is narrower than a width of each of the plurality of listening zones, second speech associated with the keyword; and

recognize the second speech to generate an indication of recognized speech.

2. The apparatus of claim 1 , wherein the instructions, when executed by the one or more processors, further cause the apparatus to:

determine, based on the indication of the recognized speech, a device;

determine, based on the indication of the recognized speech, a command; and

send the command to the device.

3. The apparatus of claim 1 , wherein the direction of the acoustic beam is fixed during the detecting the second speech.

4. The apparatus of claim 1 , wherein the direction of the acoustic beam is different than a direction of each of the plurality of listening zones.

5. The apparatus of claim 1 , wherein the instructions, when executed by the one or more processors, cause the apparatus to determine the direction by:

determining, for one or more of the plurality of listening zones, one or more scores based on the one or more indications of first speech; and

determining, based on the one or more scores, the direction.

6. The apparatus of claim 1 , wherein the plurality of microphones comprises a plurality of microphone arrays, and wherein each of the microphone arrays is associated with a different one of the plurality of listening zones.

7. The apparatus of claim 1 , wherein the plurality of listening zones comprises a plurality of acoustic beams that are each different from the acoustic beam used to detect the second speech.

8. The apparatus of claim 1 , wherein the keyword comprises a plurality of words.

9. An apparatus comprising:

one or more processors; and

memory storing instructions that, when executed by the one or more processors, cause the apparatus to:

determine, based on signals from a plurality of microphones, and for a plurality of directions, audio characteristics associated with first speech;

determine that the first speech comprises a keyword;

determine, based on the audio characteristics, a direction and a width associated with the keyword;

direct, using at least some of the plurality of microphones, an acoustic beam having the direction;

detect, using the acoustic beam, second speech; and

recognize the second speech to generate an indication of recognized speech.

10. The apparatus of claim 9 , wherein the instructions, when executed by the one or more processors, further cause the apparatus to:

determine, based on the indication of the recognized speech, a device;

determine, based on the indication of the recognized speech, a command; and

send the command to the device.

11. The apparatus of claim 9 , wherein the instructions, when executed by the one or more processors, further cause the apparatus to:

determine, for the plurality of directions, scores based on the audio characteristics; and

determine the direction based on the scores.

12. The apparatus of claim 9 , wherein the direction of the acoustic beam is different than any of the plurality of directions.

13. The apparatus of claim 9 , wherein the audio characteristics comprise one or more of amplitudes, confidence levels, or signal-to-noise ratios associated with the keyword.

14. The apparatus of claim 9 , wherein the acoustic beam has the width that is determined.

15. A non-transitory computer-readable medium storing instructions that, when executed, cause:

receiving, from a plurality of microphones, one or more indications of first speech detected in a plurality of listening zones;

determining, based on the one or more indications of the first speech, that the first speech comprises a keyword;

determining a direction associated with the keyword;

detecting, using an acoustic beam formed by at least some of the plurality of microphones, pointed in the direction associated with the keyword, and having a width that is narrower than a width of each of the plurality of listening zones, second speech associated with the keyword; and

recognizing the second speech to generate an indication of recognized speech.

16. The non-transitory computer-readable medium of claim 15 , wherein the instructions, when executed, further cause:

determining, based on the indication of the recognized speech, a device;

determining, based on the indication of the recognized speech, a command; and

sending the command to the device.

17. The non-transitory computer-readable medium of claim 15 , wherein the direction of the acoustic beam is fixed during the detecting the second speech.

18. The non-transitory computer-readable medium of claim 15 , wherein the direction of the acoustic beam is different than a direction of each of the plurality of listening zones.

19. The non-transitory computer-readable medium of claim 15 , wherein the instructions, when executed, cause determining the direction by causing:

determining, for one or more of the plurality of listening zones, one or more scores based on the one or more indications of first speech; and

determining, based on the one or more scores, the direction.

20. The non-transitory computer-readable medium of claim 15 , wherein the plurality of microphones comprises a plurality of microphone arrays, and wherein each of the microphone arrays is associated with a different one of the plurality of listening zones.

21. The non-transitory computer-readable medium of claim 15 , wherein the plurality of listening zones comprises a plurality of acoustic beams that are each different from the acoustic beam used to detect the second speech.

22. The non-transitory computer-readable medium of claim 15 , wherein the keyword comprises a plurality of words.

23. A non-transitory computer-readable medium storing instructions that, when executed, cause:

determining, based on signals from a plurality of microphones, and for a plurality of directions, audio characteristics associated with first speech;

determining that the first speech comprises a keyword;

determining, based on the audio characteristics, a direction and a width associated with the keyword;

directing, using at least some of the plurality of microphones, an acoustic beam having the direction;

detecting, using the acoustic beam, second speech; and

recognizing the second speech to generate an indication of recognized speech.

24. The non-transitory computer-readable medium of claim 23 , wherein the instructions, when executed, further cause:

determining, based on the indication of the recognized speech, a device;

determining, based on the indication of the recognized speech, a command; and

sending the command to the device.

25. The non-transitory computer-readable medium of claim 23 , wherein the instructions, when executed, further cause:

determining, for the plurality of directions, scores based on the audio characteristics; and

determining the direction based on the scores.

26. The non-transitory computer-readable medium of claim 23 , wherein the direction of the acoustic beam is different than any of the plurality of directions.

27. The non-transitory computer-readable medium of claim 23 , wherein the audio characteristics comprise one or more of amplitudes, confidence levels, or signal-to-noise ratios associated with the keyword.

28. The non-transitory computer-readable medium of claim 23 , wherein the acoustic beam has the width that is determined.