Patent · US Active

Online maximum-likelihood mean and variance normalization for speech recognition

US8996368B2 · kind B2 · utility

5Cited by

2References

24Claims

0Family size

Assignee

Nuance Communications, Inc. · US

Inventor

Daniel Willett · Eltville am Rhein, DE

Key dates

Filing date	Feb 22, 2010
Grant date	Mar 31, 2015
Priority date	—
Expiry date	Feb 1, 2031

Classification

Technology area (CPC G)Physics
CPC primaryG10L15/34
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

A feature transform for speech recognition is described. An input speech utterance is processed to produce a sequence of representative speech vectors. A time-synchronous speech recognition pass is performed using a decoding search to determine a recognition output corresponding to the speech input. The decoding search includes, for each speech vector after some first threshold number of speech vectors, estimating a feature transform based on the preceding speech vectors in the utterance and partial decoding results of the decoding search. The current speech vector is then adjusted based on the current feature transform, and the adjusted speech vector is used in a current frame of the decoding search.

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