Back to results
US 2026/0229237 A1Published application
Machine-learning techniques for dialog processing
AssigneeNVIDIA Corporation
InventorsAnshul Jain, Sumit Kumar Bhattacharya, Ratin Kumar, Jason Conrad Roche, Shubhadeep Das, Bangqi Wang, Rajath Bellipady Shetty
Life of this patent
Estimating the term…
Abstract
Apparatuses, systems, and techniques to identify speakers based on content of speech. In at least one embodiment, one or more speakers are identified based on content of speech.
Claims20
One or more processors, comprising:
- circuitry to:
- determine, from a video stream, that two or more users are speaking;
- capture an audio segment of at least one user of the two or more users;
- generate a text transcript of the audio segment;
- generate one or more indications that the text transcript corresponds to a contextual query;
- determine an intent of the text transcript;
- and
- select one or more dialog states corresponding to the at least one user based, at least in part, on the intent and the one or more indications.
5 more elements
A system comprising, one or more processors to:
- determine, from a video stream, that two or more users are speaking;
- capture an audio segment of at least one user of the two or more users;
- generate a text transcript of the audio segment;
- generate one or more indications that the text transcript corresponds to a contextual query;
- determine an intent of the text transcript;
- and
- select one or more dialog states corresponding to the at least one user based, at least in part, on the intent and the one or more indications.
4 more elements
A method comprising:
- determining, from a video stream, that two or more users are speaking;
- capturing an audio segment of at least one user of the two or more users;
- generating a text transcript of the audio segment;
- generating one or more indications that the text transcript corresponds to a contextual query;
- determining an intent of the text transcript;
- and
- selecting one or more dialog states corresponding to the at least one user based, at least in part, on the intent and the one or more indications.
4 more elements
Dependencies are inferred from the claim text.