Back to results
US 2019/0166435 A1Published application
Separating and recombining audio for intelligibility and comfort
AssigneeWhisper.ai, Inc.
InventorsDwight Crow, Shlomo Zippel, Andrew Song, Emmett McQuinn, Zachary Rich
Life of this patent
Estimating the term…
Abstract
Audio enhancement systems, devices, methods, and computer program products are disclosed. In particular embodiments, audio is separated by source at an auxiliary processing device, primary voice presence and/or relevancy per source is determined and used to determine enhancement data that is sent to one or more earpieces and used for enhancing audio at the earpiece. These and other embodiments are disclosed herein.
Claims26
1-14. (canceled)
A method of enhancing audio for at least one earpiece of a user comprising:
- receiving, at an auxiliary processing device, audio data corresponding to audio sensed at the at least one earpiece;
- receiving, at the auxiliary processing device, additional data, including non-audio data, relevant to processing to enhance the audio sensed at the at least one earpiece;
- using the received audio data to identify a plurality of respective portions of the audio data as corresponding to a plurality of respective separate audio sources;
- using the received audio data to compute audio source parameters for each respective separate audio source;
- using the received audio data and the additional data to estimate relevancy of each audio source to the user;
- using the estimated relevancy to determine volume weights for each respective separate audio source;
- and
- using the volume weights to determine and send enhancement data to the at least one earpiece for enhancing audio played for the user at the at least one earpiece.
5 more elements
A method of enhancing audio for at least one earpiece of a user comprising:
- receiving, at an auxiliary processing device, audio data corresponding to audio sensed at the at least one earpiece;
- using the received audio data to identify a plurality of respective portions of the audio data as corresponding to a plurality of respective separate audio sources;
- using the received audio data to compute audio source parameters for each respective separate audio source;
- using the computed audio source parameters to estimate whether one of the respective separate audio sources comprises a primary voice source to which the user is attending;
- and
- if one of the respective separate audio sources is estimated to comprise a primary voice source, then performing processing comprising:
- determining all respective separate audio sources that are not the primary voice source to be secondary noise sources;
- using a preferred signal-to-noise ratio corresponding to the user and a volume of the primary voice source to determine a maximum combined noise value for the secondary noise sources;
- determining and applying volume weights to each of the secondary noise sources to maintain a combined noise value for the secondary noise sources that is equal to or less than the maximum secondary noise value;
- and
- using the volume weights to determine enhancement data to send to the at least one earpiece for enhancing audio played for the user at the at least one earpiece.
8 more elements
A hearing aid system comprising:
- at least one earpiece configured to sense audio and to transmit audio data corresponding to the sensed audio;
- and
- an auxiliary processing device configured to receive the audio data from the at least one earpiece;
- to receive additional data, including non-audio data, relevant to processing to enhance the audio sensed at the at least one earpiece;
- and to perform processing comprising:
- using at least the received audio data to identify a plurality of respective portions of the audio data as corresponding to a plurality of respective separate audio sources;
- using at least the received audio data to compute audio source parameters for each respective separate audio source;
- using the received audio data and at least a portion of the additional data to estimate relevancy of each audio source to the user;
- using the estimated relevancy to determine volume weights for each respective separate audio source;
- and
- using the volume weights to determine enhancement data to send to the at least one earpiece for enhancing audio played for the user at the at least one earpiece.
8 more elements
A computer program product comprising a non-transitory computer readable medium storing executable instruction code that, when executed by a processor, performs processing comprising:
- using audio data corresponding to audio sensed by at least one earpiece to identify a plurality of respective portions of the audio data as corresponding to a plurality of respective separate audio sources;
- using the audio data to compute audio source parameters for each respective separate audio source;
- using the audio data and additional data, including non-audio data, to estimate relevancy of each audio source to the user;
- using the estimated relevancy to determine volume weights for each respective separate audio source;
- and
- using the volume weights to determine enhancement data to send to the at least one earpiece for enhancing audio played for the user at the at least one earpiece.
3 more elements
Dependencies are inferred from the claim text.