Patent · US Active

Convolutional neural network with phonetic attention for speaker verification

US11776548B2 · kind B2 · utility

1Cited by

0References

20Claims

0Family size

Assignee

MICROSOFT TECHNOLOGY LICENSING, LLC · US

Inventors

Yong Zhao · Beijing, CN
Tianyan Zhou · Bellevue, US
Jinyu Li · Beijing, CN
Yifan Gong · Sammamish, US
Jian Wu · Bellevue, US
Zhuo Chen · Markham, CA

Key dates

Filing date	Feb 7, 2022
Grant date	Oct 3, 2023
Priority date	—
Expiry date	Feb 7, 2042

Classification

Technology area (CPC G)Physics
CPC primaryG10L17/14
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

Embodiments may include determination, for each of a plurality of speech frames associated with an acoustic feature, of a phonetic feature based on the associated acoustic feature, generation of one or more two-dimensional feature maps based on the plurality of phonetic features, input of the one or more two-dimensional feature maps to a trained neural network to generate a plurality of speaker embeddings, and aggregation of the plurality of speaker embeddings into a speaker embedding based on respective weights determined for each of the plurality of speaker embeddings, wherein the speaker embedding is associated with an identity of the speaker.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.