matcha-video

Joined 2 years ago

Karma

Submissions

Member since

Oct 6

Recent submissions

https://matcha.video - In-browser "pre-editor" for cutting down a long sequence of repeated takes for talking-head content to just the stuff you want to keep. - Aligns cuts to the speaker's pose so they aren't cut-to mid-blink or mid-word

- Exports cuts to FCP and DaVinci so you can use your existing editing flow

- Transcript editor with retake grouping and word-level cuts

- Dysfluency detection ("um", "uh")

- Free, in-browser. Optional pay for faster / better server ASR

Currently in beta, free server ASR for new users. Please try it and give your honest feedback

Comment · 1 month ago

https://matcha.video - In-browser "pre-editor" for cutting down a long sequence of repeated takes for talking-head content to just the stuff you want to keep.

- Aligns cuts to the speaker's pose so they aren't cut-to mid-blink or mid-word

- Exports cuts to FCP and DaVinci so you can use your existing editing flow

- Transcript editor with retake grouping and word-level cuts

- Dysfluency detection ("um", "uh")

- Free, in-browser. Optional pay for faster / better server ASR

Currently in beta, free server ASR for new users. Please try it and give your honest feedback

Comment · 1 year ago

Same problem that got Obadiah Stane

Comment · 1 year ago

Question for folks who work a lot with STT models - What is your favorite model that supports word-level timestamps, has good dysfluency detection (whisper isn't great), and is also supported by transformers.js?

Comment · 1 year ago

Huggingface is just a git hosting service, like github. You can add up the sizes of all the files in the directory yourself