search

NVIDIA / DeepLearningExamples

State-of-the-Art Deep Learning scripts organized by models - easy to train and deploy with reproducible accuracy and performance on enterprise-grade infrastructure.
13.35k stars 3.2k forks source link

how to extend to support chinese TTS [Model/Framework or something else] Feature requested #341

Closed gaoyuqi closed 3 years ago

gaoyuqi commented 4 years ago

Related to Model/Framework(s) or something else (describe)

Examples:

Is your feature request related to a problem? Please describe. A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

Describe the solution you'd like A clear and concise description of what you want to happen.

Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered.

Additional context Add any other context or screenshots about the feature request here.

ghost commented 4 years ago

@gaoyuqi thanks for your interest in the TTS; in order to support Chinese, a relevant text cleaner is required, among other things. I'll keep this issue open for any developments to support Chinese language

ghost commented 4 years ago

Since there are a few requests about non-English datasets, we're planning to enhance our repository with the possibility to train on such datasets. Stay tuned!

lightwithshadow commented 3 years ago

Training the model on the Chinese data set, the epoch to 500 alignment effect is not ideal;

alignment_0

Except for modifying the symbols in symbols.py to the Chinese pinyin character set; Set the parameter ext-cleaners=basic_cleaners of train.py and inference.py;

are there other places that need to be modified? thx！

ghost commented 3 years ago

Hi @lightwithshadow have you checked the discussion on https://github.com/NVIDIA/tacotron2/issues/17 ?