Voice Regeneration System to Convert Whispering Voice to Pseudo-Real Voice

摘要

A software based prototype of voice regeneration system to convert whispering voice talk to pseudo-real voice talk is developed. Its intention is to allow a person to talk to the other party via phone channel just by whispering, i.e. without vibrating his or her vocal fold to generate real voice. The kernel of the signal processing system is either FFT based or correlation based vocoder to regenerate a pitched-voiced signal from whispered-unvoiced signal spectrum or autocorrelation, at its tonal part of the sequence. The system, if implemented in mobile or other style of phone, may solve major reason to prohibit its use in closed public space (such as railway wagon) due to its "acoustical nuissance" of talking voice.