Browse Prior Art Database

Voice Recognition Applied to Voice Message Systems

IP.com Disclosure Number: IPCOM000045261D
Original Publication Date: 1983-Feb-01
Included in the Prior Art Database: 2005-Feb-06
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Williams, BJ: AUTHOR

Abstract

This article describes a method of accessing a voice message store and forward system over the Public Switching Network (PSN), which does not require the use of a multi-frequency (MF) telephone set. The method is based on the use of a portable audio emitting device and a voice recognition device.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 58% of the total text.

Page 1 of 1

Voice Recognition Applied to Voice Message Systems

This article describes a method of accessing a voice message store and forward system over the Public Switching Network (PSN), which does not require the use of a multi-frequency (MF) telephone set. The method is based on the use of a portable audio emitting device and a voice recognition device.

According to the proposed method, the voice recognition device is brought into play when it is required by activating an audio emitting device (AED) capable of emitting a unique signal which lies within the telephone bandwidth and has sufficient audible power. The AED could be about the size of a cigarette lighter which contains the circuits, battery and miniature loudspeaker at one end, and a button at its other end. The button is pushed when the user wants to enter a spoken command; the audible signal is broadcast from the miniature loudspeaker at the other end of the device, and it is detected and transmitted by the microphone in the telephone handset. The tone generated by the AED could be either one selected from the Q23 MF class or a special one. In the latter case, frequency and signal duration variations (tolerances) would have to be carefully selected and monitored.

The voice recognition system would be controlled by user independent software, which in its basic form would consist of about 20 commands (the digits 0 to 9 and about 10 other words, such as LISTEN, SPEAK, etc). The user would need to train the syste...