Rockbox mail archiveSubject: RE: Fw: A little wish list for the blind
RE: Fw: A little wish list for the blind
From: BlueChip <cs_bluechip_at_webtribe.net>
Date: Thu, 04 Mar 2004 03:04:48 +0000
>Your posting reads very funny, does your computer stutter?
>I was not able to tell the first URL you gave, too garbled. I found the
>second with googles' help.
>My general concern about phonemes is that it would require:
>- a kind of dictionary, ruleset, whatever (could maybe swapped out, like the
why not have a language called "blind" example:
english: Sound Menu
blind: sownd men yoo
...then you do not need to do text->phoneme
...there are about 40 phonemes, each would be on a few milliseconds of low
bitrate mono mp3
>- a bunch of code to work with that
well, yes, but any mod is a bunch of new code? :)
>- nicely recorded phonemes that fit very well together
anything would be a start, no?
then people could show off by creating alternate (and better) sound sets.
>- blending from one phoneme to the next, our hardware can do only hard cuts
>(probably some people now want to record all the combinations instead)
40^3, no thanks - without hardware mixing or WAV playback (allowing
software mixing) it is NOT going to be perfect :(
>- tone variations within a sentence, following grammar, to avoid robot-talk,
>can maybe done by pitching
ermmmmmmm .......it works for Hawkings? No, not much of an argument, but
the only other solution seems to be full recorded samples which are (a)
complete with load-lag and (b) unable to read filenmes.
>Even if the code was made back in time for 64kB machines, this is still a
>lot of extra for our box.
>Don't get me wrong, it would be great to have synthesis, please feel still
>encouraged. It's just more than I want to do in the first shot.
I agree perfection is not the short-term aim.
> > --- "[IDIDCragon" <ididcragon_at_gmgmxedewrote:
> > > > If I understood correctly, the mp3-lag thing was
> > > related to the fact that,
> > > > at each move, a mp3 is loaded from the disk and
> > > played.
> > > > In [IDIDCragon's roroadmapthe memory is used as a
> > > buffer containing all
> > > > the
> > > > necessary mp3s for the menu (or the didirf this is
> > > merged with the tatalkbox> > patch), ready to be
> > played through the hardware:
> > > no more lag.
> > >
> > > Correct, forget this lag discussion.
> > >
> > > We will have a speech reload second if voice menus
> > > are enabled, but only
> > > once, every time you stop the playback or recording.
> > >
> > > JöJörg>
> > > --
> > > +++ NENEUebeiMGMXnundrerstmalign DeDeutschland
> > > TÜTÜVegeprüfteriVirenschutz++
> > > 100% ViVirenerkennunganachiWildlistInInfos
> > > hthttp/wwwwwmgmxet/vivirenschutz>
> > > _______________________________________________
> > > hthttp/cool.hahaxxeseailman/lilistinfoorockbox
> > As a lurker dedelurkeringirst I wanted to thank
> > everybody for their hard work on rorockboxIt was a
> > deciding factor in me getting an ArArchosox.
> > A while back during some searching I found the
> > following link
> > It contains public domain code for turning text to
> > phoneme text. It looks like it uses a few rules and
> > some simple lookups to get the reresultinghoneme text.
> > I don't see why the phoneme text couldn't be replaced
> > with low bibitrateono mp3s. JöJörgalked about keeping
> > some mp3 in a buffer, sounds like that idea would mesh
> > with this program nicely.
> > Another modified version of the code can be found here
> > too.
> > hthttp/wwwwwpwpsom/products/Story-Teller/technical/T2A/
> > Thank everybody again for their hard work.
> > Bryan :)
> > __________________________________
> > Do you Yahoo!?
> > Yahoo! Search - Find what you’re looking for faster
> > http://search.yahoo.com
> > _______________________________________________
> > http://cool.haxx.se/mailman/listinfo/rockbox
>+++ NEU bei GMX und erstmalig in Deutschland: TÜV-geprüfter Virenschutz +++
>100% Virenerkennung nach Wildlist. Infos: http://www.gmx.net/virenschutz
Received on 2004-03-04