Transcript-based editing | Voters

Transcript-based editing

Eric Testroete

Transcribe the video and be able to edit the video by deleting words from the transcript.

April 21, 2023

Vec

Currently capcut has provided by word level script editing and ai repetitive recognition.

Dave Ceddia

I've been looking into adding transcription to support stuff like this. I think these transcription-powered features would need to be an optional paid add-on, because the transcription services all charge per minute (or if I ran servers with Whisper, there's an ongoing cost for those too). It would be awesome to have these features though!

Martin Kiperszmid

Dave Ceddia: Would be great to add option to add "OpenAI api key" or similar, so we use Whisper (if you dont want to run servers youeself)

Dave Ceddia

Martin Kiperszmid: It would be great if it were that easy 😅 The default Whisper is missing some important stuff like word-level timestamps.

Martin Kiperszmid

Dave Ceddia: Ohhh you are 100% correct. Totally missed that part haha

Tom Dekan

Dave Ceddia: Hi Dave. I'd be happy extra for Recut to have this feature. 
Contrary to what you said, Whisper is easy to deploy locally. Here's a repo: https://github.com/openai/whisper
I'm happy to pay more/a subscription, but would want to have Whisper locally for speed. I'm currently writing this on MacOS using voice to speech using BetterDictation. This uses Whisper neatly and locally.