Zmanjševanje vpliva šumov in motenj pri avtomatskih sistemih telefonskega govornega dialoga /
The most complex part of an automatic telephone dialog system clearly presentsa speech recognizer. It must operate rubust, meaning that recognition rates should be invariant to unexpected variations in speech signals or in telephone lines. In order to set up a user-friendly and fatihful automatic te...
Shranjeno v:
| Main Authors: | , |
|---|---|
| Format: | Book Chapter |
| Jezik: | Slovenian |
| Teme: | |
| Sorodne knjige/članki: | Vsebovano v:
Zbornik šeste Elektrotehniške in računalniške konference ERK '97, 25. - 27. september 1997, Portorož, Slovenija |
| Oznake: |
Označite
Brez oznak, prvi označite!
|
| Izvleček: | The most complex part of an automatic telephone dialog system clearly presentsa speech recognizer. It must operate rubust, meaning that recognition rates should be invariant to unexpected variations in speech signals or in telephone lines. In order to set up a user-friendly and fatihful automatic telephone dialog system, noise and echo cancellation techniques are neccesery to be included in the feature extraction part of z recognizer. Influences of background and telephone noise, as well as distortions caused by the telphone echo signals, to the recognition system will first be analised. Methods and algorithms for noise and echo cancellation and some issues of their practical implementation will then be discussed. Algoritms will be evaluated of recognition performances simulating heavy noise and echo conditions. |
|---|---|
| Fizični opis: | Str. B/193-196. |
| ISBN: | 9616062115 |