search

NVIDIA / waveglow

A Flow-based Generative Network for Speech Synthesis
BSD 3-Clause "New" or "Revised" License
2.29k stars 530 forks source link

Is the published model trained only on LJSpeech? #234

Closed MuyangDu closed 4 years ago

MuyangDu commented 4 years ago

Is the published pretrained model trained only on LJSpeech? The filename of the published model contains 'universal' so I'm wondering if any other data besides LJSpeech was used to train. Thanks

meenalmalik commented 4 years ago

The model is trained just on LJ speech dataset. But you can surely alter/customize the length of the training and test files, by following the "train your own model" guide:)

MuyangDu commented 4 years ago

The model is trained just on LJ speech dataset. But you can surely alter/customize the length of the training and test files, by following the "train your own model" guide:)

Thanks for the explanation.