NCJ Number
57715
Date Published
1976
Length
4 pages
Annotation
THIS PAPER DESCRIBES THE COMPUTER SPEECH PROCESSING SYSTEM DEVELOPED FOR DIGITAL ENHANCEMENT OF DEGRADED VOICE RECORDINGS.
Abstract
SPEECH RECORDINGS ARE IMPORTANT AS BOTH INVESTIGATIVE AND CRIMINAL EVIDENCE. HOWEVER, THE QUALITY OF RECORDINGS OFTEN IS INADEQUATE TO SATISFY THE NEEDS OF THE FORENSIC EXAMINER. AUDIO DEGRADATIONS DUE TO BACKGROUND NOISES OR OTHER CAUSES ARE OFTEN DIFFICULT OR IMPOSSIBLE TO REMOVE WHEN USING CONVENTIONAL AUDIO FILTERING TECHNIQUES. THE COMPUTER SPEECH PROCESSING SYSTEM, WHICH CAN BE IMPLEMENTED IN A MINICOMPUTER, MAKES USE OF SEVERAL TYPES OF DIGITAL FILTERS SYNTHESIZED IN THE COMPUTER TO MINIMIZE THE EFFECTS OF ROOM ACOUSTICS, ECHOES, TELEPHONE DEGRADATIONS, MOTOR NOISE, AND TAPE RECORDER HUM. THE SYSTEM ALSO HAS THE CAPABILITY OF COMPUTER AUDIO TAPE EDITING AND DIGITAL AUTOMATIC GAIN CONTROL FOR EQUALIZING SEVERAL VOICES IN A CONVERSATION. THE SYSTEM IS BASED ON OPERATOR COMMUNICATION, CONTROL, AND OBSERVATION VIA AN INTERACTIVE AUDIOGRAPHIC CONSOLE. SEVERAL OPERATIONS CAN BE CARRIED OUT ON THE SPEECH ONCE THE SPEECH IS STORED IN THE COMPUTER. THE SPECIFIC TECHNIQUES REQUIRED FOR DIGITAL EDITING, DIGITAL FILTERING, AND ANALOG-TO-DIGITAL OR DIGITAL-TO-ANALOG CONVERSION ARE DISCUSSED. EXTENSIONS TO THE SYSTEM CURRENTLY BEING CONSIDERED INCLUDE SPEAKER SEPARATION, FILTERING OF SPEECH IN UNCORRELATED NOISE, AND CONTINUOUS SPEECH SPECTOGRAM (VOICEPRINT) GENERATION. CHARTS AND GRAPHS ILLUSTRATE THE TEXT. SEE ALSO NCJ 38864, 52569, 57707-57714, AND 57716-57726. (AUTHOR ABSTRACT MODIFIED--PRG)