	US 11,783,834 B1
	Conserving battery while detecting for human voice
Narendra Gyanchandani, Sammamish, WA (US); Bilyana Slavova, Redmond, WA (US); Njenga Kariuki, Seattle, WA (US); Matthew Heida, Snoqualmie, WA (US); Naveenan Vasudevan, Seattle, WA (US); and Stephen Phillip Pant, Seattle, WA (US)
Assigned to Amazon Technologies, Inc., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Sep. 29, 2022, as Appl. No. 17/936,713.
Application 17/936,713 is a continuation of application No. 16/862,359, filed on Apr. 29, 2020, granted, now 11,462,218.
This patent is subject to a terminal disclaimer.
Int. Cl. G10L 15/22 (2006.01); G10L 25/78 (2013.01); G06F 1/18 (2006.01); G06F 1/16 (2006.01); G06F 1/3287 (2019.01)

CPC G10L 15/22 (2013.01) [G06F 1/163 (2013.01); G06F 1/189 (2013.01); G06F 1/3287 (2013.01); G10L 25/78 (2013.01)]

20 Claims

1. A computer-implemented method, comprising:

executing, on a device that includes a battery and a plurality of battery powered components that are powered by the battery, a monitoring schedule that includes a low power state time duration and an active state time duration, wherein the plurality of battery powered components include a voice activity detector;

during a first low power state time duration of the monitoring schedule:

maintaining the plurality of battery powered components in a low power state to preserve a battery charge of the battery;

upon completion of the first low power state time duration:

transitioning the voice activity detector from the low power state to an active state;

generating, at the device and during a first active state time duration of the monitoring schedule, a first audio data; and

determining, at the device and with the voice activity detector during the first active state time duration, that a voice was not detected in the first audio data;

in response to determining that the voice was not detected in the first audio data:

transitioning the voice activity detector to the low power state;

increasing the low power state time duration of the monitoring schedule from the first low power state time duration to a second low power state time duration that is longer than the first low power state time duration; and

maintaining the plurality of battery powered components in the low power state for the second low power state time duration;

upon completion of the second low power state time duration:

transitioning the voice activity detector from the low power state to the active state;

generating, at the device and during a second active state time duration, second audio data; and

determining, at the device and with the voice activity detector during the second active state time duration, that the voice was detected in a second audio data;

in response to determining that the voice was detected:

maintaining the voice activity detector in the active state; and

continuing to collect the second audio data.