Usability

E L Tonkin py7elt at bath.ac.uk
Thu Jul 26 02:00:24 UTC 2001


Hi,

On Tue, 24 Jul 2001, Nick Hockings wrote:

> There is another side to usability: available means of input and output.
> 
> My area of interest is medical informatics, especially computer aided 
> diagnostics.
> 
> To be useful and medical computer system needs a method of input that does not 
> impede the Doctor/paramedic/nurse in any way, YET captures all pertinent 
> information.
> 
> Practically this means voice input on a palmtop.
> "pocket-linux" runs on at least two current palmtops that I know (Acompli and 
> Ipaq). However I know of no freesoftware app that can do speach-to-text, nor any 
> that can do hand-writting-to-text.
> 

AFAIK speech-to-text gives you the options of using ViaVoice or, um,
ViaVoice. But it isn't Free, nor AFAIK is it compatible. There's
KVoiceControl which, as it sounds, is a KDE desktop app... and
CVoiceControl, which uses the console. However, this is isolated word
recognition only.

> Both of these are very important modes of input for palmtops and 3rd Generation 
> cellphones.

Agreed, but definitely one of the most difficult pieces of software to
write...

> Text-to-speach is important for blind, visually impaired, and dyslexic people.
> Does anyone know of a Free app to do this? Imagine being able to write and xhtml 
>  page and use it either in a website, or a Gnome Atchung! presentation with text 
> to speach by using different XSL style sheets.

Festival:
http://www.cstr.ed.ac.uk/projects/festival/manual/festival_2.html#SEC2

It's released under a free licence. It's not however GPL - they claim that
their commercial partners preferred their version. If anybody would care
to take a look at it...? I've looked at the gnu.org licences page and it
isn't on there so far unless it's a copy of something I don't recognise.

It works ok, but needs tuning (lots of it) and takes about 30 seconds to
process one phrase on a P75... 

> I don't expect GUIs on mouse-keyboard-screen interfaces to change much. The real 
> potential for improving usability lies with new types of I/O hardware, and the 
> apps to exploit them. The "holy grail" (and a long way off) is natural language 
> interpretation and generation. For the time being, precise terminologies may 
> allow this to be faked up for specialist apps like medical informatics.

Um... depends who you intend the system for. I personally have limited use
for advanced natural language interpretation, though generation is very
useful indeed under certain circumstances. Whilst I'd appreciate the
ability to condense conversations into speech and review them later, I'd
prefer a compact keyboard/Twiddler for wearable applications, due to
the speed limitations inherent in human speech. Can see the medical uses
though.

Em





More information about the Discussion mailing list