keithito / tacotron

A TensorFlow implementation of Google's Tacotron speech synthesis with pre-trained model (unofficial)
MIT License
2.96k stars 956 forks source link

Why we use audio samples as input in this code. Research paper stated that text sequence is the input to tacotron-2 model #361

Open tanu456 opened 3 years ago

ljuvela commented 3 years ago

The model is also autoregressive on audio features.