Training seems to permanently hang on <All keys matched successfully>

[deeperinmypower]

I might be missing something obvious, but when trying to use the Train functionality, I keep running into the same issue.

I create a workspace, provide a path to a folder containing wav files of the voice I want to train a model for. Resample and split dataset seems to work fine - I keep the 'Pitch extraction method' to 'harvest', 'Filter radius' to 3, and Extract pitches. This works fine, as does Create index file.

But when I switch to train, (Base checkpoint = f0, Epochs to train = 100, Batch size 6, Save every n epochs = 10), I see my terminal either give an error:

RuntimeError: mat1 and mat2 must have the same dtype

... or the UI says 'Starting training' and the terminal hangs on:

I've tried leaving it for 30-60 minutes, but get nothing output. Am I being impatient, or is something broken?

I'm running this on an M2 Mac Mini. Happy to provide other tech specs if needed.

Thanks!

[gitmylo]

Did you use a 48k model? those aren't supported, use a 40k model

[deeperinmypower]

v1 40k gin_channels: 256 self.spk_embed_dim: 109 <All keys matched successfully> <All keys matched successfully> /code/audio-webui/venv/lib/python3.10/site-packages/torch/cuda/amp/grad_scaler.py:120: UserWarning: torch.cuda.amp.GradScaler is enabled, but CUDA is not available. Disabling. warnings.warn("torch.cuda.amp.GradScaler is enabled, but CUDA is not available. Disabling.") /code/audio-webui/venv/lib/python3.10/site-packages/torch/amp/autocast_mode.py:204: UserWarning: User provided device_type of 'cuda', but CUDA is not available. Disabling warnings.warn('User provided device_type of \'cuda\', but CUDA is not available. Disabling') Traceback (most recent call last): File "/code/audio-webui/venv/lib/python3.10/site-packages/gradio/routes.py", line 437, in run_predict output = await app.get_blocks().process_api( File "/code/audio-webui/venv/lib/python3.10/site-packages/gradio/blocks.py", line 1352, in process_api result = await self.call_function( File "/code/audio-webui/venv/lib/python3.10/site-packages/gradio/blocks.py", line 1093, in call_function prediction = await utils.async_iteration(iterator) File "/code/audio-webui/venv/lib/python3.10/site-packages/gradio/utils.py", line 341, in async_iteration return await iterator.__anext__() File "/code/audio-webui/venv/lib/python3.10/site-packages/gradio/utils.py", line 334, in __anext__ return await anyio.to_thread.run_sync( File "/code/audio-webui/venv/lib/python3.10/site-packages/anyio/to_thread.py", line 33, in run_sync return await get_asynclib().run_sync_in_worker_thread( File "/code/audio-webui/venv/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 877, in run_sync_in_worker_thread return await future File "/code/audio-webui/venv/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 807, in run result = context.run(func, *args) File "/code/audio-webui/venv/lib/python3.10/site-packages/gradio/utils.py", line 317, in run_sync_iterator_async return next(iterator) File "/code/audio-webui/webui/ui/tabs/training/training/rvc_workspace.py", line 683, in train_model ) = net_g(phone, phone_lengths, pitch, pitchf, spec, spec_lengths, sid) File "/code/audio-webui/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/code/audio-webui/venv/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1156, in forward output = self._run_ddp_forward(*inputs, **kwargs) File "/code/audio-webui/venv/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1113, in _run_ddp_forward return module_to_run(*inputs, **kwargs) File "/code/audio-webui/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/code/audio-webui/webui/modules/implementations/rvc/infer_pack/models.py", line 631, in forward o = self.dec(z_slice, pitchf, g=g) File "/code/audio-webui/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/code/audio-webui/webui/modules/implementations/rvc/infer_pack/models.py", line 495, in forward har_source, noi_source, uv = self.m_source(f0, self.upp) File "/code/audio-webui/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/code/audio-webui/webui/modules/implementations/rvc/infer_pack/models.py", line 418, in forward sine_merge = self.l_tanh(self.l_linear(sine_wavs)) File "/code/audio-webui/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/code/audio-webui/venv/lib/python3.10/site-packages/torch/nn/modules/linear.py", line 114, in forward return F.linear(input, self.weight, self.bias) RuntimeError: mat1 and mat2 must have the same dtype

I'm using a WAV file, 44.1 kHz, 16 bits per second to train the model (just in case that's useful to know :) ) - thanks for your help!

[gitmylo]

Do you have an NVIDIA graphics card? RVC training is not supported on cpu, or amd/intel gpus. Only NVIDIA.

If you don't have an NVIDIA gpu, but still want to train, you can use the google colab for that.

[deeperinmypower]

Ah, I guess as I'm using a Mac, that rules me out of running it locally - but Google Colab is an option, thanks!

One other question then - if I've made a voice clone model, is there any way to use that model in the Text to Speech interface, or does it only work for Audio to Audio RVC?

[gitmylo]

RVC is only for audio to audio, but you can use bark for text to speech, and then use that audio in RVC

[deeperinmypower]

Thanks!