THE SEMI-AUTOMATIC SYSTEM DESCRIBED IN THIS ARTICLE ANALYZES SPEECH SAMPLES TO IDENTIFY AND EXTRACT SPEAKERDEPENDENT FEATURES AND TO SUBSEQUENTLY PERFORM A STATISTICAL COMPARISON OF THE FEATURES FROM DIFFERENT SAMPLES. THE SYSTEM UTILIZES A MINI-COMPUTER AND ASSOCIATED PERIPHERALS TO ACCEPT ANALOG SPEECH SIGNALS FOR PROCESSING AND STATISTICAL COMPARISON. IN THE CRIMINAL SPEECH SAMPLE PROCESSING OPERATION, SPECIFIC PHONETIC EVENTS WHICH HAVE BEEN FOUND TO HAVE A HIGH DEGREE OF DISCRIMINATING POWER ARE IDENTIFIED AND LABELED BY THE OPERATOR USING A INTERACTIVE GRAPHIC DISPLAY TERMINAL. WHEN A SUSPECT SAMPLE IS OBTAINED, THE SAME PHONETIC EVENTS ARE SELECTED FOR PROCESSING. IN THE COMPARISON PHASE EACH SELECTED EVENT FROM THE CRIMINAL SAMPLE IS COMPARED WITH A LIKE EVENT FROM A SUSPECT SAMPLE. AFTER DESCRIBING THE OPERATIONS OF THIS SYSTEM, MEASURES OF SYSTEM PERFORMANCE AND RELIABILITY ARE DISCUSSED. IT IS NOTED THAT A MAJOR ADVANTAGE OF THIS SYSTEM IS ITS ABILITY TO UTILIZE THE BEST NATURAL SKILLS OF THE OPERATOR AND THE COMPUTER TO ARRIVE AT A NUMERICAL, REPEATABLE, OBJECTIVE CONCLUSION. (AUTHOR ABSTRACT MODIFIED)