Research on the discriminating value of different frequency formants in speaker recognition
Objectiv To study the discriminating value of different frequency formants in voice identification. Methods For-mants of 10 female and 8 male twins'natural reciting speech were measured, discrimination value and stability of which were compared and analyzed. Results whether female or male twins could be discriminated based on the formants in the different frequency ranges, including lower than 2 000Hz, 2 000Hz-5 000Hz and 5 001Hz-8 000Hz. Stability of formants were de-creased with the increasing of frequencies and the proportionf of nulli-formants was as higher as 26.67% in the range of 5000Hz-8000Hz. Conclution The discriminating value of formants within different frequency ranges was no significant differ-ence due to the influence of the stability of formants.
speaker recognitionFormant Frequencydiscriminating value
国家科技期刊平台
NSTL
万方数据
维普
