Open tralala87 opened 1 year ago
Hey Im still getting this error any way to fix??
You have to install the hubert manager. Do the following:
git clone https://github.com/gitmylo/bark-voice-cloning-HuBERT-quantizer/
cd bark-voice-cloning-HuBERT-quantizer
pip install -r requirements.txt
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu117
It should now run (This works for me in a colab notebook, I have not tested it from a cli, YMMV)
Now I get this error: ModuleNotFoundError: No module named 'bark.generation'
Oh, it's simple!
git clone https://github.com/serp-ai/bark-with-voice-clone
cd bark-with-voice-clone/
pip install git+https://github.com/suno-ai/bark.git
Now 'hubert' module will be available. :)
still getting no module named hubert
still getting no module named hubert
@vvonchain, seems like they changed name of folder. Look at https://github.com/gitmylo/bark-voice-cloning-HuBERT-quantizer/tree/master/bark_hubert_quantizer So simply change hubert to bark_hubert_quantizer
Take a look at my pull request: https://github.com/serp-ai/bark-with-voice-clone/pull/51
And here is my notebook in google colab: https://colab.research.google.com/drive/1fW6-RfcIX4wZ0T5Fc0B-M_3a459L3qaW?usp=sharing
from bark_hubert_quantizer.hubert_manager import HuBERTManager
As @BrasD99 mentioned, directory is renamed. This simple line solved my issue. Likewise, just replace all of 'hubert' to 'bark_hubert_quantizer!
still getting no module named hubert
@vvonchain, seems like they changed name of folder. Look at https://github.com/gitmylo/bark-voice-cloning-HuBERT-quantizer/tree/master/bark_hubert_quantizer So simply change hubert to bark_hubert_quantizer
Take a look at my pull request: #51
And here is my notebook in google colab: https://colab.research.google.com/drive/1fW6-RfcIX4wZ0T5Fc0B-M_3a459L3qaW?usp=sharing
Hi, I was trying to Clone an audio file using this Notebook, but the process seems to be very slow for some reason. I am using NVIDIA 4090 with 24GB vRAM
Hi, I was trying to Clone an audio file using this Notebook, but the process seems to be very slow for some reason. I am using NVIDIA 4090 with 24GB vRAM
@souvikqb Hi, I'm not a contributor to this repository. But it takes a long time for me too. You are using a gpu, so I think you need to load the appropriate models for it, so change this flags:
preload_models(
text_use_gpu=False, # -> To True
text_use_small=False,
coarse_use_gpu=False, # -> To True
coarse_use_small=False,
fine_use_gpu=False, # -> To True
fine_use_small=False,
codec_use_gpu=False, # -> To True
force_reload=False
)
Hi, I was trying to Clone an audio file using this Notebook, but the process seems to be very slow for some reason. I am using NVIDIA 4090 with 24GB vRAM
@souvikqb Hi, I'm not a contributor to this repository. But it takes a long time for me too. You are using a gpu, so I think you need to load the appropriate models for it, so change this flags:
preload_models( text_use_gpu=False, # -> To True text_use_small=False, coarse_use_gpu=False, # -> To True coarse_use_small=False, fine_use_gpu=False, # -> To True fine_use_small=False, codec_use_gpu=False, # -> To True force_reload=False )
Thank you so much, that solved it.
I was wondering if there was a way to increase the output quality somehow. ?
Currently, the cloning works, but is there any way to make it better or any parameters to play with?
Currently, the cloning works, but is there any way to make it better or any parameters to play with?
I can't help here because I've run into problems myself. Сan't even get any satisfactory results. I can’t clone the Russian voice at all, the output is random English speech. And no one answers my issue ;( But I'm glad you got the preliminary results!
UPD: i fixed my notebook.
Hi, I was trying to Clone an audio file using this Notebook, but the process seems to be very slow for some reason. I am using NVIDIA 4090 with 24GB vRAM
@souvikqb Hi, I'm not a contributor to this repository. But it takes a long time for me too. You are using a gpu, so I think you need to load the appropriate models for it, so change this flags:
preload_models( text_use_gpu=False, # -> To True text_use_small=False, coarse_use_gpu=False, # -> To True coarse_use_small=False, fine_use_gpu=False, # -> To True fine_use_small=False, codec_use_gpu=False, # -> To True force_reload=False )
Thank you so much, that solved it.
I was wondering if there was a way to increase the output quality somehow. ?
Currently, the cloning works, but is there any way to make it better or any parameters to play with?
But how can i load my new voice created with my specific voice name?
Hi, I was trying to Clone an audio file using this Notebook, but the process seems to be very slow for some reason. I am using NVIDIA 4090 with 24GB vRAM
@souvikqb Hi, I'm not a contributor to this repository. But it takes a long time for me too. You are using a gpu, so I think you need to load the appropriate models for it, so change this flags:
preload_models( text_use_gpu=False, # -> To True text_use_small=False, coarse_use_gpu=False, # -> To True coarse_use_small=False, fine_use_gpu=False, # -> To True fine_use_small=False, codec_use_gpu=False, # -> To True force_reload=False )
Thank you so much, that solved it. I was wondering if there was a way to increase the output quality somehow. ? Currently, the cloning works, but is there any way to make it better or any parameters to play with?
But how can i load my new voice created with my specific voice name?
You need to use a .npz file to save the cloned voice and then load it at the library aftwerwards.
You can find more info here: https://github.com/suno-ai/bark/issues/379
When trying to run 'clone_voice.ipynb', I get:
ModuleNotFoundError Traceback (most recent call last) in <cell line: 1>()
----> 1 from hubert.hubert_manager import HuBERTManager
2 hubert_manager = HuBERTManager()
3 hubert_manager.make_sure_hubert_installed()
4 hubert_manager.make_sure_tokenizer_installed()
ModuleNotFoundError: No module named 'hubert'