You can use a lab file to filter the audio to use or discard when you are training iVectors or audio templates. A lab file consists of time sections, each specifying a word or term - for example, you might want to use only those sections of the audio that are listed as SPEECH
.
The LabTerm
parameter specifies the word, label, or string in the lab file to keep.
For more information on speaker section labeling, see IDOL Expert.
Action: | AddTask |
Task: | IvSpkIdTrain
SpkIdTrain SpkIdTrainStream SpkIdTrainWav |
Type: | String |
Default: | * (keep all sections in the label file) |
Example: | LabTerm=SPEECH
|
See Also: | LabTerm (configuration parameter) |
|