Speaker Recognition: Basic Considerations and Methodology
Michael H. L. Hecker · The Journal of the Acoustical Society of America · 1971
This paper is based on an interpretive survey of the literature. Speaker recognition is defined as any decision-making process that uses speaker-dependent features of the speech signal. The origin and nature of these features are discussed, and three general methods of speaker recognition (aural, visual, and automatic) are described. Each method is illustrated by several experimental studies. Speaker recognition by listening appears to be the most accurate and reliable method at the present time. Speaker recognition by visual comparison of spectrograms is coming into use in criminology, but the validity of this method is still in question. Speaker recognition by machine is limited by various design shortcomings, both of concept and implementation.