A fully automatic method for text-independent speaker recognition is described. This method derives from a semiautomatic procedure which was successfully extrapolated from laboratory tests and from a number of actual applications in which the vocal signal transmitted on telephone channels were of very poor quality. The main characteristic of the procedure is the utilization of the acoustic parameters, such as formants, which have high energy in both the spectral and temporal domains, and are therefore persistent even in particularly degraded telephone signals. Since this method derives from a previous operator-assisted procedure, it is able to intervene in all feature-extraction phases, permitting control of the correctness of the parameters. Another advantage offered by the automatic system is the possibility to add automatically into the database every set accepted by the classifier, incrementing this way the statistical data.
|Publication status||Published - 1987|
|Event||Proceedings - ICASSP 87, 1987 International Conference on Acoustics, Speech, and Signal Processing. - , Unknown|
Duration: 1 Jan 1987 → …
|Conference||Proceedings - ICASSP 87, 1987 International Conference on Acoustics, Speech, and Signal Processing.|
|Period||1/1/87 → …|
All Science Journal Classification (ASJC) codes
- Signal Processing
- Electrical and Electronic Engineering
Federico, A., Ibba, G., & Paoloni, A. (1987). NEW AUTOMATED METHOD FOR RELIABLE SPEAKER IDENTIFICATION AND VERIFICATION OVER TELEPHONE CHANNELS.. Paper presented at Proceedings - ICASSP 87, 1987 International Conference on Acoustics, Speech, and Signal Processing., Unknown.