Voice detection using hearable devices
Techniques for detecting a voice command from a user of a hearable device. The hearable device may include an in-ear facing microphone to capture sound emitted from an ear of the user, and an exterior facing microphone to capture sound emitted from an exterior environment of the user. The in-ear microphone may generate an in-ear audio signal representing the sound emitted from the ear, and the exterior microphone may generate an exterior audio signal representing sound from the exterior environment. The hearable device may include components to determine correlations or similarities between the in-ear audio signal and exterior audio signal, which indicate that the audio signals represent sound emitted from the user. Further, the components may perform voice activity detection to determine that the sound emitted from the user is a voice command, and proceed to perform further voice-processing techniques.
1. An ear-based device comprising:
one or more processors;
a first microphone positioned to capture first sound from an ear of a user of the ear-based device;
a second microphone positioned to capture second sound from an environment of the user; and
computer-readable media storing computer-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
generating, using the first microphone, first audio data representing the first sound;
generating, using the second microphone, second audio data representing the second sound;
determining one or more similarities between the first audio data and the second audio data;
generating a correlated audio signal based at least in part on at least a portion of the first audio data and at least a portion of the second audio data;
processing the correlated audio signal using a voice activity detection (VAD) technique; and
determining, based at least in part on using the VAD technique to process the correlated audio signal, that the first audio data and the second audio data each represent an utterance of the user.
2. The ear-based device of claim 1 , the operations further comprising:
attenuating a first frequency range of the first audio data to generate first equalized audio data;
attenuating a second frequency range of the second audio data to generate second equalized audio data; and
analyzing the first equalized audio data and the second equalized audio data to identify the one or more similarities.
3. The ear-based device of claim 2 , wherein determining that the first audio data and the second audio data each represent the utterance of the user comprises:
identifying a portion of the second equalized audio data having the one or more similarities with the first equalized audio data; and
determining, using the VAD technique, that the portion of the second equalized audio data represents at least a portion of the utterance.
4. The ear-based device of claim 3 , wherein determining that the first audio data and the second audio data each represent the utterance of the user further comprises:
determining, using the VAD technique, that the second audio data represents the utterance.
5. The ear-based device of claim 1 , wherein determining the one or more similarities comprises:
identifying a first energy measure associated with a first frequency range of the first audio data;
identifying a second energy measure associated with a second frequency range of the second audio data, wherein the second frequency range at least partially overlaps with the first frequency range; and
identifying the one or more similarities between the first energy measure and the second energy measure.
6. The ear-based device of claim 5 , wherein determining that the first audio data and the second audio data each represent the utterance comprises:
identifying a portion of the second audio data corresponding to the second frequency range; and
determining, using the VAD technique, that the portion of the second audio data corresponding to the second frequency range represents at least a portion of the utterance.
7. The ear-based device of claim 1 , further comprising:
a loudspeaker; and
an active isolation component that at least partially acoustically isolates the first microphone from the second microphone by:
identifying a portion of the second audio data representing ambient noise from the environment of the user; and
causing the loudspeaker to emit third sound to at least partially prevent the ambient noise from being captured by the first microphone.
8. The ear-based device of claim 1 , the operations further comprising generating metadata associated with the one or more similarities between the first audio data and the second audio data, and
wherein determining that the first audio data and the second audio data each represent the utterance is further based at least in part on the metadata.