Stops very abruptly

[matthiasg]

This seems to be a pretty great model, looking forward to the next steps taken.

I do wonder however, if the abruptness in which it ends its output can be somehow mitigated ?

[DrSiemer]

It appears to crop the last part too aggressively. Adding !. at the end (and ending a sentence with punctuation before that) kind of fixes it.

[akx]

This is fixed somewhat by my PR #22.

[noname22]

The sweet spot, for me, before the fix is merged is appending "...." (that is four periods) to the string before sending it to generate. Found by trial end error, YMMV.

audio = tts.generate(text + "....", voice=voice)