The generated audio using pretrained model doesn't match the input text

[abhisheknovoic]

Hello,

Firstly, thanks for the implementation.

I used the pre-trained model and ran the synthesis.py file. I saw that the input text is "Transformer model is so fast!" while the output audio just says "Transformer model.". May I know the reason for this discrepancy ? Isn't it guaranteed to generate audio for all the input tokens?

[soobinseo]

Hi,

I also found that some inputs were not completely created and were cut off, but I couldn't figure out what was the cause. It would be good to check the attention plot of the corresponding input sequence.