Hussein Patwa wrote:
> Your mention of GSOC got me thinking. Would it be possible to make the
> plugin accessible, I mean technically, or is it a practical impossibility
> given limitations of Rockbox's architecture? Couild it be an idea to have
> this as a possible GSOC project this year? Bear in mind I know little about
> GSOC so forgive me if this is all just tripe.
Bear in mind I'm not an expert in the area myself.
My understanding is that accessibility in plugins basically requires two
steps. The first is localization of the plugins - the plugins need to
get their strings from a separate file rather than having them compiled
in. This is what the GSOC work so far has been focused on, and I believe
significant progress was made.
The second step would be actual voicing of the plugin. Since plugins are
currently able to control playback and exist concurrently with playback,
I would imagine that when plugins don't exclude playback voice would be
possible by allowing a voice file to be created off the plugin strings,
and swapping it out for the primary voice file when a plugin is loaded.
I don't know how much work this would be, and I don't mean to make it
sound easy, but merely that it seems like it should be reasonable possible.
Unfortunately plugins like mpegplayer offer an additional hurdle - they
exclude the playback engine. You can't have music playing while
mpegplayer is running. I'm not certain this will prevent voice, as the
voice file itself has its own reserved space in memory that won't be
claimed by the plugin as far as I know and voice can play when playback
is stopped, but it may require some additional effort on top of what
would be necessary for general plugin voice.
Bear in mind this is all wild speculation, and I welcome anyone more
familiar to step in and correct me. I'm just guessing, loosely speaking,
that adding voice will be significant work, and adding it in to some
specific plugins will be even more.
That being said, another alternative might be the work in integrating
espeak. If a GPLv2 version of espeak were ported, it could be added into
the plugin library and allow plugins to speak their menu structure
without pre-recorded voices. There would likely be significantly less
audio quality than some of the better pre-recorded voices but it would
The main problem with both of these approaches is that they'd require
significant programming time and time to learn the underlying systems. I
don't know how much a non-programmer can contribute to either sort of
effort, but I'm sure both ideas will be discussed as we approach this
year's GSOC as accessibility improvements are a popular idea to post to
help students think of proposals. Unfortunately simply having good ideas
does not mean the students who read them will find them as interesting
as you would find them useful.
Received on 2010-01-31