Method and apparatus for recognizing speech by using wavelet tra

Details Method and apparatus for recognizing speech by using wavelet tra Method and apparatus for recognizing speech by using wavelet tra

1993-11-08

1996-06-18

MacDonald, Allen R.

395 231, 395 257, G10L 506

Patent

active

055287255

ABSTRACT:
The present invention implements a method of automatic speech recognition in a personal computer environment and in real time. The invention uses a wavelet transformer to decompose digitized input speech templates into coefficients assigned to fundamental building blocks, from which the original speech function can be reconstructed. In the preferred embodiment of the present invention, an energy vector comprising seven wavelet coefficients can represent adequately any input speech utterance. The wavelet transformer reduces greatly the number of Input speech samples needed for storage in the computer memory. Computational economy is further achieved when the present invention quantizes the plurality of energy vectors into two binary bits with reference to a predetermined parameter. In the present invention, the predetermined parameter is established as the mean of the energy vector. At the same time, the transient response of the energy vectors are also obtained. With the transient response, the present invention aligns the input speech templates with that of the reference templates in the library without having to time wrap the time axis of the respective templates. The present invention checks if the distance between the transient response of a test template and that of a reference template exceeds a predetermined threshold.

REFERENCES:
patent: Re33597 (1991-05-01), Levinson et al.
patent: 4370521 (1983-01-01), Johnston et al.
patent: 4790017 (1988-12-01), Hiaton
patent: 4815134 (1989-03-01), Picone et al.
patent: 4905288 (1990-02-01), Gerson et al.
patent: 4989248 (1991-01-01), Schalk et al.
patent: 5014134 (1991-05-01), Lawton et al.
patent: 5384725 (1995-01-01), Coifman et al.
Synthesis/Coding of Audio Signals Using Optimized Wave lets Sinha et al IEEE 23-26 Mar. 1992.
High-Fidelity Audio Compression: Fractional-Band Wavelets Heegard et al IEEE 23-26 Mar. 1992.
Detection of a Transient Signal of Unknown Scaling and Arrival Time Using the Discrete Wavelet Transform Frisch et al IEEE/May 1991.
Detection of Transients Using Discrete Wavelet Transform Petropulu IEEE/Mar. 1992.
A Neural Net Acoustic Phonetic Feature Extractor Based on Wavelets Davenport et al IEEE/May 1991.

Affiliated with

Also associated with

Rating

Method and apparatus for recognizing speech by using wavelet tra does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Method and apparatus for recognizing speech by using wavelet tra, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for recognizing speech by using wavelet tra will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-229932