Patent · US Expired

Perceptual harmonic cepstral coefficients as the front-end for speech recognition

US7337107B2 · kind B2 · utility

8Cited by

8References

16Claims

0Family size

Assignee

The Regents of the University of California · US

Inventors

Kenneth Rose · Goleta, US
Liang Gu · Dongguan, CN

Key dates

Filing date	Oct 2, 2001
Grant date	Feb 26, 2008
Priority date	—
Expiry date	Mar 28, 2024

Classification

Technology area (CPC G)Physics
CPC primaryG10L2025/935
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

Pitch estimation and classification into voiced, unvoiced and transitional speech were performed by a spectro-temporal auto-correlation technique. A peak picking formula was then employed. A weighting function was then applied to the power spectrum. The harmonics weighted power spectrum underwent mel-scaled band-pass filtering, and the log-energy of the filter's output was discrete cosine transformed to produce cepstral coefficients. A within-filter cubic-root amplitude compression was applied to reduce amplitude variation without compromise of the gain invariance properties.

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