search

neonbjb / tortoise-tts

A multi-voice TTS system trained with an emphasis on quality
Apache License 2.0
13.22k stars 1.83k forks source link

ModuleNotFoundError: No module named 'tortoise.models' #384

Closed Bananamax closed 1 year ago

Bananamax commented 1 year ago

Greetings ladies and gentlemen.

Well I followed the instructions and installed Anaconda and git and pytorch and installed the other stuff via

git clone https://github.com/neonbjb/tortoise-tts.git cd tortoise-tts python -m pip install -r ./requirements.txt python setup.py install conda install -c conda-forge pysoundfile

and then after generating an audio file with "python tortoise/do_tts.py --text "I'm going to speak this" --voice random --preset fast" it says:

Traceback (most recent call last): File "C:\ProgramData\anaconda3\tortoise-tts\tortoise\do_tts.py", line 7, in from api import TextToSpeech, MODELS_DIR File "C:\ProgramData\anaconda3\tortoise-tts\tortoise\api.py", line 12, in from tortoise.models.classifier import AudioMiniEncoderWithClassifierHead ModuleNotFoundError: No module named 'tortoise.models'

So what now?

Bananamax commented 1 year ago

I am totally fucked here nothing works

neonbjb commented 1 year ago

It sounds like python setup.py install probably failed. It's looking for the packages from this repo and can't find them. You can also try running your command from the tortoise directory.

Bananamax commented 1 year ago

Well I did the "python setup.py install" command again and there are no errors as far as I can see. And doing "python do_tts.py --text "I'm going to speak this" --voice random --preset fast" directly in the tortoise directory also doesnt change anything. The same tortoise-models error occurs :(

Lets say I want to reinstall it all try again fresh what are the necessary steps here? Uninstall anaconda and python and also probably delete the set "PATH" and then restart it all or isnt that enough and stuff gets mixed and more errors might occur?

And also are there more strict requirements for the current Windows 11 version like a specific python version only works? As far as I know it needs to be 3.10 or lower but otherwise it doesnt matter? Using this site here with the code provided should always work? https://pytorch.org/get-started/locally/ The "solving environment" process for example sometimes takes a long time and sometimes doesnt complete at all. So to me this smells like for Windows 11 not all python and maybe cuda versions work at all. Maybe conda is better maybe pip....unclear

bbergeron0 commented 1 year ago

If that can assist you, I found it relevant to share my setup in the other issue (https://github.com/neonbjb/tortoise-tts/issues/281#issuecomment-1498111332).

Bananamax commented 1 year ago

Yeah thanks for that. So you are not using anything from here: https://pytorch.org/get-started/locally/ What about cuda? I will try your stuff as well but so far it looks like no matter what I do and reinstall it ends with the error: ModuleNotFoundError: No module named 'tortoise.models' No idea how to reset that properly now and get back on track....dubious

Edit: No it is once again generating an audio file might work :) But apparently its all on the CPU now so you are not using Cuda? How could I implement cuda in your "model"? :) I will probably try this: pip install torch==1.13.1+cu117 torchvision==0.14.1+cu117 torchaudio==0.13.1 --extra-index-url https://download.pytorch.org/whl/cu117

And this part here: Create and activate a new python environment named 'venv' using python 3.9 (Other options: 3.8, 3.7) python3.9 -m venv venv && . venv/bin/activate

The command python3.9 doesnt exist its just python right? I had to do it in 2 steps: python -m venv venv and python -m venv/bin/activate

Man its all working incl. cuda thank you very much and god bless you bbergeron0

bbergeron0 commented 1 year ago

Bless you too. Don't forget to close the issue once everything's going smoothly!