Patent · US Expired

Nonlinear mapping for feature extraction in automatic speech recognition

US7254538B1 · kind B1 · utility

26Cited by
2References
20Claims
0Family size

Assignee

Inventors

Key dates

Filing dateNov 16, 2000
Grant dateAug 7, 2007
Priority date
Expiry dateAug 16, 2024

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG10L15/144
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

The present invention successfully combines neural-net discriminative feature processing with Gaussian-mixture distribution modeling (GMM). By training one or more neural networks to generate subword probability posteriors, then using transformations of these estimates as the base features for a conventionally-trained Gaussian-mixture based system, substantial error rate reductions may be achieved. The present invention effectively has two acoustic models in tandem—first a neural net and then a GMM. By using a variety of combination schemes available for connectionist models, various systems based upon multiple features streams can be constructed with even greater error rate reductions.

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