Re: [g-a-devel] Happy patch bonanza



>>>>> "EZ" == Enrico Zini <enrico enricozini org> writes:

    EZ> Is it ok for it to be defined straight into
    EZ> /usr/share/festival/voices/italian/pc_diphone/festvox/pc_diphone.scm,
    EZ> added at the end like this?

    EZ>   (proclaim_voice
    EZ>    'pc_diphone
    EZ>    '((language italian)
    EZ>      (gender male)
    EZ>      (dialect none)
    EZ>      (description
    EZ>       "This voice provides a Italian male voice using a
    EZ>        residual excited LPC diphone synthesis method.  The lexicon
    EZ>        is provived by a set of letter to sound rules producing pronunciation
    EZ>        accents and syllabification.  The durations, intonation and
    EZ>        prosodic phrasing are minimal but are acceptable for simple
    EZ>        examples.")
    EZ>      (coding "ISO-8859-1")))

Yes, except that it's probably better to specify the coding without
double quotes: (coding ISO-8859-1)

    EZ> Because if like this is ok, I can upload the new package in
    EZ> Debian and prod the Ubuntu people to pick it up.  Not that it
    EZ> would make a big difference since ISO-8859-1 would be a default,
    EZ> but it would set an example for others to follow.

Good idea, thanks!

    EZ> BTW, is "coding" a name that comes from some standard or a name that
    EZ> we're inventing right now?  

I introduced this name in festival-freebsoft-utils.  AFAIK it's not used
anywhere else and the only voice that declares it is our Czech voice.

    EZ> Because if we're inventing it right now, then I think I'd prefer
    EZ> "encoding".

Well, I'm not sure which of the two English terms better fits the
context.

    EZ> (list 'VOICEDESC: 
    EZ> (nth 0 (voice.description '%s))
    EZ> (nth 1 (nth 0 (nth 1 (voice.description '%s))))
    EZ> (nth 1 (nth 2 (nth 1 (voice.description '%s))))
    EZ> (nth 1 (nth 1 (nth 1 (voice.description '%s)))))

    EZ> And I think I can fix this as well:

    EZ> (list 'VOICEDESC:
    EZ>  (nth 0 (voice.description '%s))
    EZ>  (cadr (assoc 'language (cadr (voice.description '%s))))
    EZ>  (cadr (assoc 'dialect  (cadr (voice.description '%s))))
    EZ>  (cadr (assoc 'gender   (cadr (voice.description '%s))))
    EZ>  (or
    EZ>    (cadr (assoc 'coding (cadr (voice.description '%s))))
    EZ>    "ISO-8859-1"))

Yes, this is the right thing to do.

Regards,

Milan Zamazal





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]