SpeechBias

If you set SpeechBias to True, an acoustic segment is declared to be speech unless there is strong evidence that it is music, noise, or silence. If you set SpeechBias to False, the winner is picked neutrally.

This parameter enables you to filter out unclear data (by setting SpeechBias to False) in cases where, for example, recognized transcripts are available. This prevents messy or inaccurate transcriptions of music or noise.

Alternatively, in cases where the recognized transcripts are used for search applications but are otherwise not read, for optimal performance you should set SpeechBias to True to recognize everything.

Type: Boolean
Default: True
Required: No
Configuration Section: audiopreproc module
Example: SpeechBias=True
See Also: SpeechBias (action parameter)

_HP_HTML5_bannerTitle.htm