by Jun OGATA and Masataka GOTO
Speech Repair is a novel
speech input interface function in which recognition errors can be
easily corrected by selecting candidates.
During the speech
input, the speech-repair function successively displays not only the usual word
sequence as a recognition result but also other competitive candidates that
indicate alternative word hypotheses obtained in the decoding process of the
speech recognizer.
Since candidates for almost all words in the recognition
result are always automatically displayed on a screen, the user can correct
erroneous words merely by selecting the appropriate candidate even while
speaking to the recognizer.
The speech-repair function is thus efficient and
easy to use: errors can be handled by simply clicking automatically displayed
candidates without waiting for the end of an utterance.
Video clips
References
J. Ogata and M. Goto,
``Speech Repair: Quick Error Correction Just by Using Selection Operation
for Speech Input Interfaces''
Proc. Interspeech 2005, pp.133-136, 2005. [PDF]