research diary
edim 2002
Week 7:
Testing aural style sheets and Installing Emacspeak
With the free trail of the speakthis Web site is now possible to test some basic styles. Some styles are not covered by the speakthis Web site.
Properties not supported
Currently the volume, spell-out, pause, cue, mixing, spatial, pitch-range, stress, richness, speak-punctuation and speak-numeral properties are not supported by Fonix SpeakThis. We are planning to add support for some of these properties in the future.
We will try adding the following styles to basic Web page.
speak: normal;
voice-family: usenglishfemale;
speech-rate: slow;
pitch: low;
}
H2 {
speak: normal;
voice-family: usenglishfemale;
speech-rate: medium;
pitch: medium;
}
H3 {
speak: normal;
speech-rate: fast;
pitch: high;
}
As you can see speakthis pastes a link into the web page which sends the
html to the speakthis site. A mp3 file is generated and sent back to the
users site, hopefully with styles added. The link did not work on Internet
Explorer 5 on a Mac (probably a plug in missing) but did work on Netscape
6.2. The audio rendered however the styles were not applied. A email was
sent to fonix asking why the styles didn’t work.
Installing Emacspeak.
As stated in week 6 it was decided to attempt to
install Emacspeak on a PC. Installing Emacspeak is a four-step process
1. Partition Hard drive
2. Install Linux
3. Install Emacspeak
4. Install Via Voice files
See installing
Emacspeak
All the software is free but it is a rather complicated process with huge
files to download (1.5 gig). Two other problems arose. One Emacspeak relies
on ViaVioce files and the files where not on the IBM web site as stated
in the "How to install Emacspeak" document. It took hours searching
the net to find the correct Linux compatible files. Second problem is that
Linux is not compatible with all sound cards. I could not find out what
sound card is on the SGI machine that I am using, fingers crossed it is
the right one.
The elusive Via Voice were files finally tracked down on a mail group
http://www.cs.vassar.edu/~priestdo/emacspeak/msg00338.html
The direct link to the files are:
ftp://people.redhat.com/jlamb
Emacspeak may be a wonderful aid for the vision impaired but how many Vision
Impaired people are running Linux with the correct sound card? How many
have the resources to partition their hard drive and install 1.5 gig of
software in the correct order to create a dual operating system?
From my research I believe the long-term solution for audio narration would be to either create a Visual browser that incorporates text to voice software. Or develop a plugin that works with Netscape and Internet Explorer.