Second International Conference on Spoken Language Processing (ICSLP'92)
Banff, Alberta, Canada
We describe a new real-time isolated word recognizer. The recognizer is designed for an extension number guidance system which looks up and announces an extension number by voice input over the telephone network. To deal with telephone quality speech input which includes noise and distortion during transmission over the telephone network, we have developed feature extraction and a word detection algorithm. These techniques use wide band-pass filter outputs which are generally used to decide whether speech is voiced sound or unvoiced sound. The recognizer is implemented with six DSP chips and executes the above functions in real-time. Finally, the recognizer is evaluated using a large telephone voice database consisting of more than 500 speakers.
Bibliographic reference. Kuroiwa, Shingo / Takeda, Kazuya / Yato, Fumihiro / Yamamoto, Seiichi / Owa, Kunihiko / Shozakai, Makoto / Matsumoto, Ryuji (1992): "Architecture and algorithms of a real-time word recognizer for telephone input", In ICSLP-1992, 1523-1526.