Back to results
US 11,636,860 B2Granted
Word-level blind diarization of recorded calls with arbitrary number of speakers
AssigneeVerint Systems Ltd.
InventorsAlex Gorodetski, Oana Sidi, Ron Wein, Ido Shapira
Life of this patent
Estimating the term…
Abstract
Disclosed herein are methods of diarizing audio data using first-pass blind diarization and second-pass blind diarization that generate speaker statistical models, wherein the first pass-blind diarization is on a per-frame basis and the second pass-blind diarization is on a per-word basis, and methods of creating acoustic signatures for a common speaker based only on the statistical models of the speakers in each audio session.
Claims20
A method of blind diarization of audio data having a first-pass blind diarization process and a second-pass blind diarization process, the method comprising:
- identifying non-speech segments in audio data and segmenting the audio data into a plurality of utterances that are separated by the identified non-speech segments;
- extracting acoustic features from the audio data;
- generating a Gaussian Mixture Model (GMM) using the extracted acoustic features;
- for each utterance of the plurality of utterances, constructing an utterance model from the GMM for the utterance;
- determining distances between each utterance of the plurality of utterances using the utterance models;
- clustering the utterances of the plurality of utterances using the determined distances;
- constructing a plurality of speaker models from the clustered utterances;
- classifying the plurality of speaker models to identify a set of common speaker GMMs and a set of generic speaker GMMs;
- constructing a first super GMM from the set of common speaker GMMs, wherein the first super GMM captures the set of common speaker GMMs;
- constructing a second super GMM from the set of generic speaker GMMs, wherein the second super GMM captures the set of generic speaker GMMs;
- receiving additional audio data;
- and
- identifying a common speaker of the additional audio data using the first and second super GMMs.
10 more elements
A system comprising:
- at least one processor;
- and
- at least one non-transitory computer-readable storage medium storing instructions that, when executed by the at least one processor, cause the system to:
- identify non-speech segments in audio data and segmenting the audio data into a plurality of utterances that are separated by the identified non-speech segments;
- extract acoustic features from the audio data;
- generate a Gaussian Mixture Model (GMM) using the extracted acoustic features;
- for each utterance of the plurality of utterances, construct an utterance model from the GMM for the utterance;
- determine distances between each utterance of the plurality of utterances using the utterance models;
- cluster the utterances of the plurality of utterances using the determined distances;
- construct a plurality of speaker models from the clustered utterances;
- classify the plurality of speaker models to identify a set of common speaker GMMs and a set of generic speaker GMMs;
- construct a first super GMM from the set of common speaker GMMs, wherein the first super GSM captures the set of common speaker GMMs;
- construct a second super GMM from the set of generic speaker GMMs, wherein the second super GMM captures the set of generic speaker GMMs;
- receive additional audio data;
- and
- identify a common speaker of the additional audio data using the first and second super GMMs.
13 more elements
A non-transitory computer-readable storage medium storing instructions that, when executed by at least one processor, cause a system to:
- identify non-speech segments in audio data and segmenting the audio data into a plurality of utterances that are separated by the identified non-speech segments;
- extract acoustic features from the audio data;
- generate a Gaussian Mixture Model (GMM) using the extracted acoustic features;
- for each utterance of the plurality of utterances, construct an utterance model from the GMM for the utterance;
- determining distances between each utterance of the plurality of utterances using the utterance models;
- cluster the utterances of the plurality of utterances using the determined distances;
- construct a plurality of speaker models from the clustered utterances;
- classify the plurality of speaker models to identify a set of common speaker GMMs and a set of generic speaker GMMs;
- construct a first super GMM from the set of common speaker GMMs, wherein the first super GMM captures the set of common speaker GMMs;
- construct a second super GMM from the set of generic speaker GMMs, wherein the second super GMM captures the set of generic speaker GMMs;
- receive additional audio data;
- and
- identify a common speaker of the additional audio data using the first and second super GMMs.
10 more elements
Dependencies are inferred from the claim text.