Forensic Speaker Recognition

This edited collection gathers research from 35 speaker recognition specialists working in many countries, among them Switzerland, Sweden, Italy, France, Japan, India and the United States. Its contributors come from speech engineering and signal processing, and together they examine how science can decide whether a suspect's voice matches recordings tied to a terrorist attack or other crimes. Those recordings are gathered by police and counter-terrorism agencies.

The chapters approach the subject from several angles. They cover the difficulties of casework in forensic settings, the problem of speech signals that have degraded, and the features of speaker recognition that can be tuned to improve voice verification systems. Other chapters look at designing voice applications that fit what law enforcement and counter-terrorism agencies actually need. A recurring idea runs through the material: the demands of forensic use push the field toward higher standards in every part of speaker recognition.

The book is intended for forensic speech scientists, speech signal processing experts and developers of speech systems. It will also interest criminal prosecutors and counter-terrorism intelligence officers and agents.