Podcastle Ideas

I have noticed that filler words like "um" and "uh" usually start about 3ms before the transcription says they do. When clicking a word in the transcript, you should highlight the range for that word on the wavform and allow a user to set a custom In and Out for that word. Then allow deleting directly from the timeline to ensure there is no lingering "u" sound in the edited audio. Understanding that audio transcription models are not 100% accurate, this would be a useful feature to ensure clean audio. Also note, that the wavform is not regenerated when removing words or spaces from the transcript. This would need to be fixed before or at the same time as implementing this feature.