I was able to create a new file and add my main video file and its .wav audio file as well as four separated .wav audio files for each participant in the video. Then, I was able to use the silence recognizer to detect silences on each participant’s .wav file (and then their speech segments went into a separate channel). This worked really well!
Is it possible to change the annotations after the recognizer has finished? I think the answer is no from what I’ve read, but I’d like to merge the recognized utterances into one annotation (if they are all one turn in the conversation). Or would it be best to just create another tier and hand-annotate those selections?
I’m not sure if I understand the question well, but a tier created by a recognizer can be edited like any other tier. When it comes to merging there are the options (for the active annotation) to merge with next or previous annotation (the annotation after or before ‘this’ one). When a turn has been segmented into multiple utterances, it would require multiple merge actions; there’s no function to merge more than two annotations at a time.
Changing the settings for the recognizer can make it to produce less segments, but then you might have to manually split annotations later, if multiple turns have been recognized as one utterance (which may not be so likely if, as in your case, you have separate recordings per participant).
Thank you! Merging is what I was looking for - that works really well! I was trying to do it before putting the annotations into a tier - oops. Thanks for your help and sorry for my beginner questions - just getting used to the program.
