Jakub Piotr Cłapa

I think I've seen this error when I had one shard containing just a single sample (in two files) and I was doing infinite reshuffling. I think it randomly reselected...

Hey, I think it would be nice to merge this to make the tool a bit easier to install for people?

@signalprime Sure, I'll see what I can do :)

@signalprime Btw. do you have a Discord? Maybe we could have a chat there?

The "Profiler function" warning sohuld be 100% harmless. I am not sure where forking may be used during inference. Were you able to figure out the cause of the issue?

Manual listening tests with MOS seems to be the only reliable metric right now. Could be an interesting community project to make a leaderboard for TTS models with crowdsourced scoring.

I am not getting the error you are seeing with these samples. They are not spoken correctly but the model finished generating successfully. Would you mind trying to find a...

It may be worth looking into the recent [ExecuTorch](https://pytorch.org/executorch/stable/intro-how-it-works.html) as a possible alternative. WhisperSpeech already works with `torch.compile` but I think we had some issues with `torch.export` around our implementation...

Hey, good question :) The `8` is `sqrt(64)` which is the default `head_width` (so by default, the whole expression is going to be `query_mult * 1`). Similar to the `3`...

Hey, you need to download all the models manually and everything else should work. Here is an example that pulls the models into a Docker image: https://github.com/collabora/WhisperFusion/blob/c58a8d1446a768833811aca9dbdfb6388957fdcb/docker/scripts/setup-whisperfusion.sh#L19-L23