Currently ONNXRuntime's CoreML support is not very good, and on my environment, I get errors like “coreml_execution_provider.cc:192 operator() Input (/sdp/flows.3/GatherND_2_output_0) has a dynamic shape ({-1,-1}) but the runtime shape ({0,10}) has zero elements. This is not supported by the CoreML EP.” and inference fails.
I hope this will be fixed in a future version of ONNXRuntime, and leave the test code as it is.
ONNXRuntime has relied on CUDA 12.x / cuDNN 9.x by default since v1.19.0, while PyTorch finally supports cuDNN 9 since v2.4.0.
Therefore, unless the pinning of torch<2.4 is eliminated, CUDA cannot be used with the latest version of ONNXRuntime.
librosa: Originally not used under style_bert_vits2/.
pyannote.audio: Required only when using the "infer style vector from audio" function during inference, but it is currently almost unused.
scipy: Because it was used in only one utility function that was not needed outside of training.
Prebuilt wheels for almost every OS/architecture (except musl) are now available on PyPI, eliminating the need for a build environment.
ref: https://pypi.org/project/pyworld-prebuilt/
Style-Bert-VITS2 has been reported to not work with some PyTorch 2.2 series, but Python 3.12 is only supported in Torch >= 2.2, so Python 3.12 support is not provided for the time being
