keithito / tacotron

A TensorFlow implementation of Google's Tacotron speech synthesis with pre-trained model (unofficial)
MIT License
2.96k stars 957 forks source link

Mask memory with zeros for values past the encoder sequence lengths #213

Closed npuichigo closed 6 years ago

npuichigo commented 6 years ago

I think maybe it's more appropriate to mask the padding values of the encoder memory to make alignments easier to learn.