Dúvida sobre esse modelo e o outro modelo large-pt-v2

#1
by mrwikrom - opened

I hope you're doing well. I was wondering if you could help me with a question I have regarding the "whisper-large-v2-pt-v3" and "whisper-large-v2-pt" models.

I noticed that the "whisper-large-v2-pt-v3" model has a Wer of '4.8385', which is lower than the Wer of '5.590' achieved by the "whisper-large-v2-pt" model. Does this mean that the "whisper-large-v2-pt-v3" model has outperformed the "whisper-large-v2-pt" model?

I bring this up because the "whisper-large-v2-pt" model is currently the best performing model in terms of evaluation, according to:
https://paperswithcode.com/sota/automatic-speech-recognition-on-mozilla-67

However, the "whisper-large-v2-pt-v3" model has a lower Wer score but is not ranked first. I would appreciate any insight you can offer to help me understand why this is the case.

Thank you very much for your help.

Best regards,
Lucas Rodrigues

Interesting, I was just looking if there was a Brazilian portuguese ASR model and found this, gonna try both version, thanks!

mrwikrom changed discussion status to closed

Sign up or log in to comment