Subject: Re: festival (voice synth. package)
To: Frederick Bruckman <fb@enteract.com>
From: Richard Rauch <rauch@eecs.ukans.edu>
List: netbsd-help
Date: 05/27/2001 09:25:56
> I built mbrola and mbrolavox-us1 just yesterday, from current pkgsrc.
> The only thing that _didn't_ get built yesterday is "festival-doc".
Hm. Interesting. Okay...
> > > It is kind of complicated to set up. If you have festival-doc
> > > installed to the usual place, the user guide is at
> >
> > Maybe I didn't do enough in setting it up. As I read the info pages, it
> > looks like most of the configuration is done at compile-time. The default
> > behavior seemed to be set up ``mostly sanely'', so I didn't fiddle much
> > with that. I'll see if it might be fixed by festival configuration,
> > though.
>
> I was responding to the "no default" error. Clearly, nas is ruled out,
Maybe we're talking at cross-purposes, here. The only explicit problem
I've had with defaults is that if I only have the aec (diphone) voice
installed, festival claims that it can't find any voices. If I have us1
(mbrola) installed, then I can at least get festival more or less
running.
The problem that has my attention is that mbrola seems to dump core with
an FPE (I get an mbrola.core if I'm in a directory that I can write).
The only way around this is to use a non-mbrola voice (even though it
seems like I may need to have an mbrola voice available just to get
festival to play ball).
I'll see if I can sort anything out by tweaking config files.
"I probably don't know what I'm talking about." --rauch@eecs.ukans.edu