Wakeword detection
Techniques for processing incoming audio using multiple wakeword detectors are described. Audio data representing an utterance may be processed by different wakeword detectors that can detect different wakewords and are associated with different speech processing components. When a wakeword is detected by one of the wakeword detectors, it may be processed by the corresponding speech processing component.
1. A computer-implemented method, comprising:
receiving first audio data corresponding to an utterance;
processing the first audio data using:
a first wakeword detection component configured to detect a representation of at least a first wakeword, the first wakeword detection component being associated with a first speech processing component, and
a second wakeword detection component configured to detect a representation of at least a second wakeword different from the first wakeword, the second wakeword detection component being associated with a second speech processing component;
determining the utterance includes the first wakeword; and
causing the first speech processing component to process further audio data.
2. The computer-implemented method of claim 1 , wherein:
the first speech processing component is associated with a first speech processing system;
the second speech processing component is associated with a second speech processing system different from the first speech processing system; and
the method further comprises sending the further audio data to the first speech processing component associated with the first speech processing system.
3. The computer-implemented method of claim 1 , wherein:
the first wakeword detection component is associated with a first companion application; and
the second wakeword detection component is associated with a second companion application different from the first companion application.
4. The computer-implemented method of claim 1 , wherein:
the first wakeword is associated with a first skill; and
the second wakeword is associated with a second skill different from the first skill.
5. The computer-implemented method of claim 1 , further comprising:
determining, using the first wakeword detection component, a first confidence value corresponding to the first wakeword; and
determining, using the second wakeword detection component, a second confidence value corresponding to the second wakeword,
wherein determining the utterance includes the first wakeword is based at least in part on the first confidence value and the second confidence value.
6. The computer-implemented method of claim 1 , wherein operation of the first wakeword detection component occurs at least partially in parallel with operation of the second wakeword detection component.
7. The computer-implemented method of claim 1 , wherein the first wakeword detection component comprises at least one trained model trained with respect to a user's voice.
8. A system comprising:
at least one processor; and
at least one memory comprising instructions that, when executed by the at least one processor, cause the system to:
receive first audio data corresponding to an utterance;
process the first audio data using:
a first wakeword detection component configured to detect a representation of at least a first wakeword, the first wakeword detection component being associated with a first speech processing component, and
a second wakeword detection component configured to detect a representation of at least a second wakeword different from the first wakeword, the second wakeword detection component being associated with a second speech processing component;
determine the utterance includes the first wakeword; and
cause the first speech processing component to process further audio data.
9. The system of claim 8 , wherein:
the first speech processing component is associated with a first speech processing system;
the second speech processing component is associated with a second speech processing system different from the first speech processing system; and
the at least one memory further comprises instructions that, when executed by the at least one processor, further cause the system to send the further audio data to the first speech processing component associated with the first speech processing system.
10. The system of claim 8 , wherein:
the first wakeword detection component is associated with a first companion application; and
the second wakeword detection component is associated with a second companion application different from the first companion application.
11. The system of claim 8 , wherein:
the first wakeword is associated with a first skill; and
the second wakeword is associated with a second skill different from the first skill.
12. The system of claim 8 , wherein the at least one memory further comprises instructions that, when executed by the at least one processor, further cause the system to:
determine, using the first wakeword detection component, a first confidence value corresponding to the first wakeword; and
determine, using the second wakeword detection component, a second confidence value corresponding to the second wakeword,
wherein the instructions that cause the system to determine that the utterance includes the first wakeword are based at least in part on the first confidence value and the second confidence value.
13. The system of claim 8 , wherein operation of the first wakeword detection component occurs at least partially in parallel with operation of the second wakeword detection component.
14. A computer-implemented method, comprising:
receiving first audio data corresponding to an utterance;
processing the first audio data using a first wakeword detection component to determine that the utterance includes at least a first wakeword;
at least partially in response to determining that the utterance includes the first wakeword, using a second wakeword detection component different from the first wakeword detection component to determine that the utterance includes at least a second wakeword; and
based at least in part on determining that the utterance includes the second wakeword, causing a speech processing component associated with the second wakeword to process further audio data.
15. The computer-implemented method of claim 14 , further comprising:
at least partially in response to determining that the utterance includes the first wakeword, using a third wakeword detection component different from the first wakeword detection component and the second wakeword detection component to determine if the utterance includes at least a third wakeword different from the second wakeword.
16. The computer-implemented method of claim 15 , wherein operation of the second wakeword detection component occurs at least partially in parallel with operation of the third wakeword detection component.
17. The computer-implemented method of claim 15 , wherein:
the second wakeword detection component is associated with a first speech processing system;
the third wakeword detection component is associated with a second speech processing system different from the first speech processing system; and
the method further comprises sending the further audio data to the speech processing component associated with the first speech processing system.
18. The computer-implemented method of claim 15 , wherein:
the second wakeword detection component is associated with a first companion application; and
the third wakeword detection component is associated with a second companion application different from the first companion application.
19. The computer-implemented method of claim 15 , wherein:
the second wakeword is associated with a first skill; and
the third wakeword is associated with a second skill different from the first skill.
20. The computer-implemented method of claim 14 , wherein the first wakeword detection component is associated with a hardware component of a device that detected the utterance.