Running speech on long sentences.

[vijayendra-tripathi]

I recently tried TTS and results are very exciting. However I noticed that speech breaks if a sentence gets longer. The terminal shows message Decoder stopped with max_decoder_steps 500 Is there a way I can make the model run on longer sentences without retraining the model.

[Vagif12]

I'm also trying to figure this out

[erogol]

you can change max_decoder_stepsin the model config which is where the model is downloaded in your system.

[Vagif12]

This is it, works fine now!

[Vagif12]

@erogol in the sample demo(https://erogol.github.io/ddc-samples/) for the sentence:

Bill got in the habit of asking himself “Is that thought true?” and if he wasn’t absolutely certain it was he just let it go.

The wordwasn'tis pronounced correctly.
However, after modifying the max_decoder_steps in the config file of the tacotron config file, and using the above sentence with the tts command like so:
tts --text "Bill got in the habit of asking himself “Is that thought true?” and if he wasn’t absolutely certain it was he just let it go."

The word 'wasn't' is not pronounced correctly, and the intonation is different to the one seen on the demo. Any reason for this behaviour?
Thanks!

P.S - when given 2 sentences with full stops, there is a "gap" between the end of the first sentence and the beginning of second sentence. Any idea on how to solve this?