SURVEY OF SPEECH DIGITIZATION TECHNIQUES (FROM CARNAHAN CONFERENCE ON ELECTRONIC CRIME COUNTERMEASURES, 1972 PROCEEDINGS, BY J S JACKSON AND R W DEVORE - SEE NCJ-10711)

• NCJ Number: 27918
• Author(s): J W BAYLESS; S J CAMPANELLA; A J GOLDBERG
• Date Published: 1972
• Length: 13 pages
• Corporate Author: University of Kentucky; Institute of Electrical and Electronics Engineers
• Publication Format: Document
• Publication Type: Best Practice/State-of-the-Art Review
• Language: English
• Country: United States of America

Annotation

A TUTORIAL, HIGHLY-TECHNICAL SURVEY OF THE STATE-OF-THEART IN DIGITAL ENCODING OF SPEECH SIGNALS.

Abstract

PRESENTLY, VOICE DIGITIZATION TERMINALS CAN PRODUCE INTELLIGIBLE SPEECH AT DATA RATES ABOVE 1200 BPS. AT BIT RATES MUCH BELOW 20 KBPS, THESE TECHNIQUES GENERALLY INVOLVE COMPLEX PROCESSING AND WITH FEW EXCEPTIONS PRODUCE SPEECH THAT SOUNDS SYNTHETIC. AT DATA RATES ABOVE 20 KBPS, THE VOICE DIGITIZATION EQUIPMENT BECOMES GREATLY SIMPLIFIED AND PROVIDES NATURAL SOUNDING SPEECH. LINEAR PREDICTIVE ENCODING OFFERS THE PROMISE OF NATURAL SOUNDING SPEECH AT BIT RATES BELOW 10 KBPS; ALTHOUGH AT PRESENT, THIS TECHNIQUE EXISTS ONLY AS A COMPUTER SIMULATION AND HAS NOT BEEN REDUCED TO HARDWARE. OVER THE NEXT DECADE, MODERN CIRCUIT DESIGN TECHNIQUES PROMISE TO REDUCE THE COST AND SIZE OF THE VOICE DIGITIZATION HARDWARE TO THE POINT WHERE COMPLEX PROCESSING IN THE VOICE TERMINAL CAN BECOME A PRACTICAL ALTERNATIVE PROVIDING NATURAL SOUNDING SPEECH BELOW 10 KBPS. A BIBLIOGRAPHY IS INCLUDED. (AUTHOR ABSTRACT)