Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Visualise transcription confidence with sharded color highlights to speed up transcript review #87

Open
menelic opened this issue Oct 9, 2024 · 1 comment

Comments

@menelic
Copy link

menelic commented Oct 9, 2024

Whisper produces confidence estimates for chunks, ie words or phrases, of the transcribed text. Please visualise these in the editor by producing html text highlights, for example in different gradients of white through yellow to dark orange. Such a visual aide would greatly speed up transcription review.

thanks a lot for this great tool!

@kaixxx
Copy link
Owner

kaixxx commented Oct 9, 2024

I did implement this in the very first versions of noScribe that used MS Word as an editor. I have since moved away from Word and ditched the support for these confidence markers in the process. Besides some technical difficulties, the main reason was that these markers weren't very helpful to begin with.
Here is an example transcribed with noScribe 0.3. Everything below and up to a confidence level of 3 is marked in red:

grafik

I found that checking all these small red bits one by one is so much work that we might as well go through the whole file once while listening to the original audio. I've tried to make this as easy as possible in the noScribe editor. You can even speed up the audio if you want. This is also much safer, since sections with a high confidence level might still contain errors.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants