synthesised wave are silent and 0:00 length

keithito / tacotron

A TensorFlow implementation of Google's Tacotron speech synthesis with pre-trained model (unofficial)

MIT License

2.95k stars 959 forks source link

synthesised wave are silent and 0:00 length #280

Open derejehinsermu opened 5 years ago

derejehinsermu commented 5 years ago

I have trained 200K steps on Afaan Oromoo(ethiopian Local language) dataset and the output at the checkpoints are good but synthesized audio is silent, i used basic_cleaners in Hparameters 1000 max-iterations and batch_size 2 step-200000-align step-198000-align Additionally why is my alignment graph full purple color, thanks

mligema commented 5 years ago

https://github.com/keithito/tacotron/issues/92

kashifibnaejaz commented 2 years ago

i too get like this could anyone solve this problem?