Keep filler "umms" and "uhhs" in the transcript?

[pmorrison1989]

I'm hoping someone can help me...

I want to keep filler "umms" and "uhhs" in the transcript, so that I can import the transcript into Davinci Resolve and use the native edit-by-text feature to edit those filler words out of the video.

Is there a way/mode to use Buzz so that the transcript contains ALL of the "umm"s and "uhh"s?

[raivisdejus]

@pmorrison1989 Buzz does not filter these sounds out, we display whatever underlying Whisper models give us. You can search for options to keep these sounds in some whisper engine or model, the community may have ideas.

One such idea I found here - openai/whisper#1174

My experiments with initial prompts has not resulted in any meaningful differences in the transcription output, but this may be case that works. Buzz supports "initial prompt" in the Advanced preferences. See if it works for you and let us know. If this works i would like to update our documentation with this example