Rockbox mail archiveSubject: Re: Playing order in playlists
Re: Playing order in playlists
From: David Pedersen <dav-ped_at_frisurf.no>
Date: Sun, 18 Dec 2005 22:19:58 -0500
Does ALL the voiced items have to sit in memory buffer "permanently"? OK,
the way things are done right now. But how about if we let the voice ui read
the text to speak from disk. That way, when it is to say 'a', it will go
into the voice.lang file, pick the right spot, and read/say whatever is in
that spot. Yes, it may slow down operation somehow, and limit battery life
on each charge. But we would gain the benefit of making all characters &
menu choices voiced, and the software even better for those of us, who do
not see everything.
----- Original Message -----
From: "Jens Arnold" <arnold-j_at_t-online.de>
Sent: Sunday, December 18, 2005 3:49 PM
Subject: Re: Playing order in playlists
> On 18.12.2005, Jani Kinnunen wrote:
>>> The virtual keyboard has all ISO8859-1 characters, at least
>>> on my player. I don't know how it is voiced, though.
>> They aren't voiced at all. I examined the .lang files, the
>> English and Finnish ones, and noticed that there aren't
>> specified any other voiced letters than the usual A to Z. Then
>> I tried to add, for example, the following lines at the end of
>> the finnish.lang file:
>> id: VOICE_CHAR_Ä
>> desc: spoken only, for spelling
>> eng: ""
>> voice: "Ä"
>> new: ""
> This cannot work. There has to be support for the additional
> characters in the code. Only the unaccented latin letters, the
> numbers and the dot are voiced, no other characters. If we want
> voice support for other characters, this does potentially mean
> thousands of them, especially as we support full unicode now.
> This isn't possible with the current approach. The voice files
> are really tight already, on archos because of the total RAM
> size limit, and on iriver because the buffer for the voice file
> is allocated permanently, and takes RAM away from the audio
> Regards, Jens
Received on 2005-12-19