Python and pytorch version ?

[Michaelwhite34]

AppData\Local\Temp\pip-install-r50u5l8a\praat-parselmouth_4b359ba8e3b34072ab1c1d1b3e3b0992\setup.py", line 44, in run raise RuntimeError("CMake must be installed to build the following extensions: " + ", ".join(e.name for e in self.extensions)) RuntimeError: CMake must be installed to build the following extensions: parselmouth [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip. error: legacy-install-failure

× Encountered error while trying to install package. ╰─> praat-parselmouth

note: This is an issue with the package mentioned above, not pip. hint: See above for output from the failure.

[Michaelwhite34]

On ubuntu, python 3.8. I used conda to install scipy and pip to install other dependences.

(prodiff) michael@michael-Legion-Y7000P-2019:~/Downloads/ProDiff$ CUDA_VISIBLE_DEVICES=$GPU python inference/ProDiff.py --config modules/ProDiff/config/prodiff.yaml --exp_name ProDiff --hparams="N=4,text='the invention of movable metal letters in the middle of the fifteenth century may justly be considered as the invention of the art of printing.'" --reset Traceback (most recent call last): File "inference/ProDiff.py", line 2, in from inference.base_tts_infer import BaseTTSInfer ModuleNotFoundError: No module named 'inference'

[Rongjiehuang]

Please run with export PYTHONPATH=.

Indeed, it would be a little bit inconvenient to install several packages. Please try and deal with the conflict (if any).

[Michaelwhite34]

When should I run the line? Before test ? Or before all dependency installation?

[keelezibel]

@Michaelwhite34 you should install the dependencies first. During execution of the python scripts whether inference or others, you should run export PYTHONPATH=. so that the corresponding python scripts will check the current working directory for the paths.

[Michaelwhite34]

Now I get this

python data_gen/tts/bin/pre_align.py --config egs/datasets/audio/1/base_text2mel.yaml | Hparams chains: ['egs/datasets/audio/1/base_text2mel.yaml'] | Hparams: binarization_args: {'shuffle': True}, binary_data_dir: data/binary/1, debug: False, exp_name: , infer: False, mel_loss: ssim:0.5|l1:0.5, min_frames: 128, num_spk: 2320, num_test_samples: 30, pitch_type: frame, pre_align_cls: egs.datasets.audio.libritts.pre_align.LibrittsPreAlign, processed_data_dir: data/processed/1, raw_data_dir: data/raw/1, test_num: 200, use_spk_id: True, validate: False, vocoder_ckpt: , work_dir: , Traceback (most recent call last): File "data_gen/tts/bin/pre_align.py", line 20, in pre_align() File "data_gen/tts/bin/pre_align.py", line 14, in pre_align process_cls = getattr(importlib.import_module(pkg), cls_name) File "/home/michael/anaconda3/envs/prodiff/lib/python3.8/importlib/init.py", line 127, in import_module return _bootstrap._gcd_import(name[level:], package, level) File "", line 1014, in _gcd_import File "", line 991, in _find_and_load File "", line 975, in _find_and_load_unlocked File "", line 671, in _load_unlocked File "", line 843, in exec_module File "", line 219, in _call_with_frames_removed File "/home/michael/Downloads/ProDiff/egs/datasets/audio/libritts/pre_align.py", line 3, in from data_gen.tts.base_pre_align import BasePreAlign ModuleNotFoundError: No module named 'data_gen.tts.base_pre_align'

[keelezibel]

@Michaelwhite34 this repo is not complete. If you check the folders, base_pre_align.py is not there. You will have to refer to the other repo for the pre-processing scripts. @Rongjiehuang it will be nice to include the pre-processing as well, I still have not figured out the pre-processing to be able to reproduce the same training outputs.

[keelezibel]

I have tried running the preproc with Natspeech repository as mentioned in the README. There are few minor changes but overall didnt encounter much issues. The processed dataset can be trained with the existing scripts in this repo. I am finetuning the Teacher with a different voice atm, let's see how it goes.