search

keonlee9420 / Comprehensive-Transformer-TTS

A Non-Autoregressive Transformer based Text-to-Speech, supporting a family of SOTA transformers with supervised and unsupervised duration modelings. This project grows with the research community, aiming to achieve the ultimate TTS
MIT License
318 stars 41 forks source link

An errors with running the preprocess.py #13

Open dasstyxx opened 2 years ago

dasstyxx commented 2 years ago

I'm trying to preprocess the VCTK dataset, and stuck on the 'Computing statistic quantities' step. When I copy from repo preprocessed_data files instead, the training run successful.

Firstly, there is a runtime error:

preprocessor.py

625: cont_lf0_lpf_norm = (cont_lf0_lpf - logf0s_mean_org) / logf0s_std_org
RuntimeWarning: invalid value encountered in true_divide

After applying a simple crutch to fix a value of logf0s_std_org, next error appear:

165: energy_mean = energy_scaler.mean_[0]
'StandardScaler' object has no attribute 'mean_'

win 10 conda python 3.6.15 all packages from the requirements.txt is installed

dmzubr commented 2 years ago

I'm trying to preprocess the VCTK dataset, and stuck on the 'Computing statistic quantities' step. When I copy from repo preprocessed_data files instead, the training run successful.

Firstly, there is a runtime error:

preprocessor.py

625: cont_lf0_lpf_norm = (cont_lf0_lpf - logf0s_mean_org) / logf0s_std_org
RuntimeWarning: invalid value encountered in true_divide

After applying a simple crutch to fix a value of logf0s_std_org, next error appear:

165: energy_mean = energy_scaler.mean_[0]
'StandardScaler' object has no attribute 'mean_'

win 10 conda python 3.6.15 all packages from the requirements.txt is installed

Faced the same problem.

What type of duration modeling do you use? Supervised or unsupervised?

It seems that preprocess pipeline contains actions for both.So - it missed some supervised (MFA outputs) artifacts.

dasstyxx commented 2 years ago

I'm trying to preprocess the VCTK dataset, and stuck on the 'Computing statistic quantities' step. When I copy from repo preprocessed_data files instead, the training run successful. Firstly, there is a runtime error: preprocessor.py

625: cont_lf0_lpf_norm = (cont_lf0_lpf - logf0s_mean_org) / logf0s_std_org
RuntimeWarning: invalid value encountered in true_divide

After applying a simple crutch to fix a value of logf0s_std_org, next error appear:

165: energy_mean = energy_scaler.mean_[0]
'StandardScaler' object has no attribute 'mean_'

win 10 conda python 3.6.15 all packages from the requirements.txt is installed

Faced the same problem.

What type of duration modeling do you use? Supervised or unsupervised?

  • config/model.yaml -> duration_modeling.learn_alignment

It seems that preprocess pipeline contains actions for both.So - it missed some supervised (MFA outputs) artifacts.

Unsupervised + liu2021

Yeah, I forgot to mention compute_f0_stats method, which produce an error when len(f0s) == 0. It's a bit weird.

God damnit, so I must use MFA anyway. Will try, thank you!