Meeting-adapted language model for speech recognition
AssigneeMicrosoft Technology Licensing, LLC
InventorsZiad Al Bawab, Anand U Desai, Shuangyu Chang, Amit K Agarwal, Zoltan Romocsa, Christopher H Basoglu, Nathan E Wohlgemuth
Life of this patent
Estimating the term…
Abstract
A system includes acquisition of meeting data associated with a meeting, determination of a plurality of meeting participants based on the acquired meeting data, acquisition of e-mail data associated with each of the plurality of meeting participants, generation of a meeting language model based on the acquired e-mail data and the meeting data, and transcription of audio associated with the meeting based on the meeting language model.
Claims19
A system comprising:
- a memory storing processor-executable instructions;
- a processing unit to execute the processor-executable instructions to configure the system to:
- receive a request to transcribe audio associated with a meeting, the request including a meeting identifier corresponding with meeting data, the meeting data including at least meeting participants;
- use the meeting data to determine availability of a meeting language model that is associated with the meeting participants;
- and
- generate a meeting transcription of the meeting deterministically with either a generic language model or, alternatively, an adapted language model that is created from the generic language model and the meeting language model, based on the determined availability of the meeting language model and by at least:
- (i) determining that the meeting language model is unavailable;
- and
- generating the meeting transcription of the meeting with the generic language model in response to determining that the meeting language model is unavailable;
- or
- (ii) determining that the meeting language model is available;
- generating an adapted language model based on the meeting language model and the generic language model;
- and
- generating the meeting transcription with the adapted language model.
A method for generating a transcript of a meeting, the method comprising:
- receiving a request to transcribe audio associated with a meeting, the request including a meeting identifier corresponding with meeting data, the meeting data including at least meeting participants;
- using the meeting data to determine availability of a meeting language model that is associated with the meeting participants;
- and
- generating a meeting transcription of the meeting deterministically with either a generic language model or, alternatively, an adapted language model that is created from the generic language model and the meeting language model, based on the determined availability of the meeting language model and by at least:
- (iii) determining that the meeting language model is unavailable;
- and
- generating the meeting transcription of the meeting with the generic language model in response to determining that the meeting language model is unavailable;
- or
- (iv) determining that the meeting language model is available;
- generating an adapted language model based on the meeting language model and the generic language model;
- and
- generating the meeting transcription with the adapted language model.
A method for generating a transcript of a meeting, the method comprising:
- receiving a request to transcribe audio associated with a meeting, the request including a meeting identifier corresponding with meeting data, the meeting data including at least meeting participants;
- using the meeting data to determine availability of a plurality of meeting language models that are associated with the meeting participants;
- selecting a set of one or more of the plurality of meeting language models to use for generating an adapted language model;
- generating the adapted language model based on both (1) a generic language model and (2) the selected set of one or more of the plurality of meeting language models;
- and
- generating a meeting transcription of the meeting with the adapted language model.
Dependencies are inferred from the claim text.