Rockbox mail archiveSubject: Re: Voices
From: [IDC]Dragon <idc-dragon_at_gmx.de>
Date: Mon, 30 Jan 2006 09:26:01 +0100 (MET)
> I'd be willing to do the edit work.
> I have a recording studio and spend quite a bit of time editing anyway; I
> really don't mind it at all.
That was actually my first idea of voice files, when speccing the feature:
"just" record them. A bit later I learned how well TTS works these days and
that it's much easier to maintain with automated and scripted authoring.
Since we now have quite a lot of clips (353 as of today), the "key" to an
efficient voice file is to have the clips as short as possible. This is also
necessary to have numbers and spelling fluent. I don't know what audio tools
are available for such clipping and leaving the first syllable intact.
You'd need to save the clips into 353 files then, each named like the
original language ID. The authoring tool can then process them. Perhaps we
can come up with a better way, not requiring this tedious intermediate step.
A hypothetical tool could auto-clip the takes if spoken with a gap, save
them as individual files. (And hope for no offset within)
-- DSL-Aktion wegen großer Nachfrage bis 28.2.2006 verlängert: GMX DSL-Flatrate 1 Jahr kostenlos* http://www.gmx.net/de/go/dslReceived on 2006-01-30