With all the integration into EveryVoice, stand-alone installation of this repo is likely to no longer be possible.
Currently, the pyproject.toml file lists all the dependencies as if you could clone and install this, but in practice that doesn't work at the moment (primarily because you get EV 0.4 since 0.5 is not published yet).
Are we going to want to maintain the ability to clone StyleTTS2 and install it? I think the answer should be no because that would be hard to maintain. I think the install-from-source procedure should be clone EV, init the submodules, and pip install -e . in EV. And if you actually want the styletts2 CLI, subsequently cd into the submodule and run pip install -e . there.
Otherwise we're signing up for a maintenance nightmare trying to keep the submodule's main branch compatible with the latest version of EV published to PyPI.
With all the integration into EveryVoice, stand-alone installation of this repo is likely to no longer be possible.
Currently, the
pyproject.tomlfile lists all the dependencies as if you could clone and install this, but in practice that doesn't work at the moment (primarily because you get EV 0.4 since 0.5 is not published yet).Are we going to want to maintain the ability to clone StyleTTS2 and install it? I think the answer should be no because that would be hard to maintain. I think the install-from-source procedure should be clone EV, init the submodules, and
pip install -e .in EV. And if you actually want thestyletts2CLI, subsequently cd into the submodule and runpip install -e .there.Otherwise we're signing up for a maintenance nightmare trying to keep the submodule's main branch compatible with the latest version of EV published to PyPI.