Re: cxxpiper

[Kyle]

Derek Davies amesema:
> I'm having trouble finding multi-speaker models that fill in the speaker
> map with sensible values.  I would guess the value would be some human
> readable useful description, such as "female" or "nora", but they all
> appear to be gibberish.  I tried about three model sources.


I found a similar problem using Pied. The speaker is just not filled in,
so one only hears speaker 0. It does however tell me how many speakers a
voice has, so I can edit the speech-dispatcher module configuration it
puts in my user home and add -speaker x on the generated command line.
It's pretty bad though for some voices, as LibriTTS for example has well
over 900 speakers, so I have to play with random numbers to find a
speaker I like.

~Kyle