Browse Prior Art Database

Interpretation Of Speech Sound Sequences

IP.com Disclosure Number: IPCOM000097900D
Original Publication Date: 1961-Sep-01
Included in the Prior Art Database: 2005-Mar-07
Document File: 1 page(s) / 11K

Publishing Venue

IBM

Related People

Bakis, R: AUTHOR

Abstract

In many devices for interpreting speech sound sequences, such as speech recognition systems, the sound sequences that correspond to each reference word are stored as a dictionary.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 100% of the total text.

Page 1 of 1

Interpretation Of Speech Sound Sequences

In many devices for interpreting speech sound sequences, such as speech recognition systems, the sound sequences that correspond to each reference word are stored as a dictionary.

The dictionary entries for each reference word indicate those sounds that should occur in a particular sequence, although not necessarily contiguously, and those sounds that should not occur.

The entries may also specify maximum and minimum sound durations. Since there are several, or many, acceptable sound sequences for each word, the dictionary may be reduced in size by storing sequences of groups of sounds that should or should not occur. For example, assume a reference word contains the following acceptable sound sequences: Sound 1 not followed by Sound 4 followed by Sound 6

Sound 2 not followed by Sound 4 followed by Sound 6

Sound 3 not followed by Sound 4 followed by Sound 6

Sound 1 not followed by Sound 5 followed by Sound 6

Sound 2 not followed by Sound 5 followed by Sound 6

Sound 3 not followed by Sound 5 followed by Sound 6

The dictionary may be reduced in size by storing: Sounds 1 or 2 or 3, not followed by Sounds 4 or 5, followed by Sound 6. Thus, the storing of all sequences of the reference word requires eighteen memory positions; the storing of the reduced sequences requires only six positions.

1