search

open-mmlab / Amphion

Amphion (/æmˈfaɪən/) is a toolkit for Audio, Music, and Speech Generation. Its purpose is to support reproducible research and help junior researchers and engineers get started in the field of audio, music, and speech generation research and development.
https://openhlt.github.io/amphion/
MIT License
4.41k stars 373 forks source link

[Help]: Data preprocessing for NaturalSpeech2 TTS #179

Open AvivSham opened 4 months ago

AvivSham commented 4 months ago

Hi All, How are you? Thank you for your wonderful work maintaining Amphion. We are trying to train NS2 model using LibriTTS dataset, however, the steps for preprocessing the data are not clear. For example, there are two scripts - libritts.py (which is under preprocessors dir) and preprocess.py (under tts dir), we assumed that we should first run libritts.py followed by preprocess.py. We were able to create most of the files required for training, but we still missed the code file required for ns2_dataset.py i.e. code = np.load(self.utt2code_path[utt]).

In general, it is not clear what preprocessing steps should be followed before running the train.py file. Can you please write down clear steps to follow?

Thanks in advance. @netagl

casafurix commented 4 months ago

Having the same problem, any help would be amazing, thanks

a897456 commented 4 months ago

same question +1

AvivNavon commented 4 months ago

Having the same issue +1

CreepJoye commented 3 months ago

Same question+1,does anyone train NS2 successfully? Ask for help