pre trained model works but goes crazy on some sentences which are a bit long

[Tetsujinfr]

hi
first thanks for the great repo and pretty good inference performance!
I have just ran the pretrained voice model on the test text sentences.
I know that some text to speech models are quite sensitive to long inferences and can go crazy.

I attach the inference, of the text below, that is pretty creepy to listen to.

Is it just me or others got the same artefacts? Is it just a matter of training the model longer and/or tweak the hyper params or is it a known limitation of deepvoice 3 capabilities?

Inferred text:

Once upon a time there was a dear little girl who was loved by every one who looked at her, but most of all by her grandmother, and there was nothing that she would not have given to the child.

1_20180505_deepvoice3_checkpoint_step000640000.zip