thank you @julien-blanchon, it looks like tests are failing. Could you please take a look at it?

Hey, just show your commit. Sorry I missed this message. Is it okay ?

I'm running some test. I will keep up updated @julien-blanchon!

thanks @julien-blanchon for this PR... since this model is using huggingface transformer, it should be moved to huggingface_transformer folder similar to Encodec.

@julien-blanchon Thanks for this PR. I have reviewed it, and everything works fine. I only made minor modifications, such as moving the model to the huggingface_transformer folder.

We recently released a benchmark for discrete audio and speech (DASB). You can find detailed information in this paper. The code is available in this repository. Since you have already added a new tokenizer to SpeechBrain, it would be a great contribution if you could also participate in the benchmark and evaluate this tokenizer across various tasks defined in DASB. I would be more than happy to help you through the process.

Hey @poonehmousavi that a super good idea, I'm in. See you there

Hummm, I'm quite unfamiliar with this codebase, I will need to read your paper before. I've open a Draft for us to discuss: speechbrain/benchmarks#40

@poonehmousavi what should we do here? If you are happy with the code, please merge it. I don't like that, again, we add something untested. Please do verify that it is really impossible to test!