Hi HN! In the past few years with the TalkBank project, I've been hacking on a tool to make transcript analysis dead simple with CMU's TalkBank Project.
What's transcript analysis? Everything from ASR (Whisper!), morphology and syntax (grammar, your sentences, etc.), captioning and alignment (you have a transcript and you have audio, you want to know when the transcript happens line by line), or disfluencies (detect stuttering, etc.)
What's transcript analysis? Everything from ASR (Whisper!), morphology and syntax (grammar, your sentences, etc.), captioning and alignment (you have a transcript and you have audio, you want to know when the transcript happens line by line), or disfluencies (detect stuttering, etc.)
We've also trained custom models specifically for these tasks, such as a Whisper ASR model that's better for understanding disfluencies (https://huggingface.co/talkbank/CHATWhisper-en) and for segmenting text utterances for captioning and analysis (https://huggingface.co/talkbank/CHATUtterance-en).
I hope this can be helpful for the community for both speech science and beyond.