Speech recognition utilizing multitude of speech features

US7464031B2 · kind B2 · utility

Assignee

• International Business Machines Corporation · US

Inventors

• Scott E. Axelrod · Campion Road, US
• Sreeram Viswanath Balakrishnan · New Delhi, IN
• Stanley Chen · Yorktown Heights, US
• Yuging Gao · Mount Kisco, US
• Ramesh Gopinath · Millwood, US
• Hong-Kwang Kuo · Pleasantville, US
• Benoit Maison · White Plains, US
• David Nahamoo · Great Neck, US
• Michael A. Picheny · White Plains, US
• George Andrei Saon · Stamford, US
• Geoffrey G. Zweig · Oakland, US

Key dates

• Filing date: Nov 28, 2003
• Grant date: Dec 9, 2008
• Expiry date: Apr 6, 2026

Classification

• Technology area (CPC G): Physics
• CPC primary: G10L2015/085
• WIPO field: Computer technology
• WIPO sector: Electrical engineering

Abstract

In a speech recognition system, the combination of a log-linear model with a multitude of speech features is provided to recognize unknown speech utterances. The speech recognition system models the posterior probability of linguistic units relevant to speech recognition using a log-linear model. The posterior model captures the probability of the linguistic unit given the observed speech features and the parameters of the posterior model. The posterior model may be determined using the probability of the word sequence hypotheses given a multitude of speech features. Log-linear models are used with features derived from sparse or incomplete data. The speech features that are utilized may include asynchronous, overlapping, and statistically non-independent speech features. Not all features used in training need to appear in testing/recognition.