Performance / WER overview of all model options

[Appfinity-development]

Is there a comprehensive overview that shows the difference between the various model option the faster whisper library has.

Eg distil-large-v3 vs large-v3 vs large-v3-turbo

I know performance also heavily depends on model configuration. But still some general recommendations would be great from the authors (@MahmoudAshraf97).

OpenAI did publish an overview when releasing their 'large-v3-turbo' in October:

openai/whisper#2363

But it still hard to decide what would give best results for your own use case.

[Purfview]

what would give best results

large-v2

[MahmoudAshraf97]

This is the only comparison I have that contain all models
#1101 (comment)
But you still have to benchmark on your own use case to find the best WER/speed tradeoff