Rockbox mail archive
Subject: Re: voice build change?
From: [IDC]Dragon (idc-dragon_at_gmx.de)
> > Sorry to be such a nag today, but I don't think this is viable, because
> > then
> > the voice files are not downward compatible. If you come with a new
> > software
> > which has extra strings, it would fetch into the first voice-only
> > entries.
> How would it do that? The first voice-only entry starts at number 0x8001.
> You'd have to get a quite a large amount of new strings before you'd risk
Ah, OK, I understood you that your delimiter is just floating ontop of the
LANG_ IDs. If it's in safe distance, we have the two-table approach.
> > So, requests to speak a new ID which is not present just gives silence.
> > this new separation into LANG_/VOICE_ we need two "directory" tables in
> > voice file.
> No, but you need to handle a (fairly large) gap in the index numbers. But
> you would already handle gaps, don't you?
Not in this way. Currently, the ID is directly used as an offset into the
directory. It is just checked against an upper bound, which is part of the
voice file header. Non-voiced IDs have a zero directory entry. This wastes 8
bytes per such entry, on the other hand saves storing the ID itself, but the
voice file is huge anyway. I'm not searching for an ID. I could either do that,
or split into two tables.
Either way, I agree to your approach of sorting the voice-only IDs into a
way higher range. This will overcome my resistance to introduce a lot more of
them, e.g. for spelling, file types, etc.
+++ NEU bei GMX und erstmalig in Deutschland: TÜV-geprüfter Virenschutz +++
100% Virenerkennung nach Wildlist. Infos: http://www.gmx.net/virenschutz
Page was last modified "Jan 10 2012" The Rockbox Crew