公开(公告)号：JPH1138992A

公开(公告)日：1999-02-12

申请号：JP15292398

申请日：1998-06-02

Applicant: IBM

Inventor： KANEVSKIJ DMITRIJ ,  STEPHEN HERMAN MACE ,  PETER S PUUN ,  CARL PUROCHIRO

IPC: G10L11/00 ,  G10L15/00 ,  G10L15/02 ,  G10L15/10 ,  G10L15/20 ,  G10L15/28 ,  G10L17/00 ,  G10L21/02 ,  H04M3/42 ,  G10L3/00 ,  G10L3/02

Abstract: PROBLEM TO BE SOLVED: To permit an accurate speech recognition communication between a portable sound signal preprocessor(SSP) device and an automatic speech/ speaker recognition(ASSR) by converting feature vector data produced from digital signals into acoustic signals and communicating through a communication channel. SOLUTION: A ciphering device 130 ciphers data preprocessed by a digital signal processor(DSP) 120, and guarantees a safe exchange of the signals through a telephone line. A compression device 140 compresses the preprocessed data before transfer, and increases the data in the quantity forwarded and transmitted by audio signals when a band width of the line is narrow. And, the preprocessing of digitized data from a microphone 10 generates a sequence of a feature vector representing a speech pattern spoken to the microphone 100. Further, a receiver part of an acoustic coupler 150 converts audio signals from an ear pad of the receiver into digital signals to process them with DSP 120.