Whisper by OpenAI

Did anyone give this a go as other users are quite impressed with it, I no longer have a Pi4 as joined the Rock5b fan club :slight_smile:
Anyone got a Pi4 to bench and test output results?

Running on a Rock5a results very impressive, still though I had to max out the input and put AGC on as often the input level of soundcards and headsets on linux is awful.
Give Whisper a clear good recorded signal and results are far better than anything else you will see.

I am still undecided if a streaming ASR mode is needed at all with short command sentences.

j1nx from OVOS as if my telepathy posted some results.

PS has anyone tried the bigger models of Whisper as for some reason it not mentioned but you can have crazy SNR levels and the medium or large model compared to others by magic get transcription correct.

I have been playing with DeepFilterNet which is great but not much good for Whisper as it actually makes results worse.
If you have a model that you can train then likely you would have to preprocess the dataset with DeepFilterNet but quite likely in terms of noise do as well. OpenAi have kept the training of Whisper to themselves but you should try the larger models with noise as the results are outstanding.

1 Like