Dear All
I'm glad Barry has started this line of discussion. I would be very interested if people could share their experience of speech recognition environmental control systems, specifically:
- how frequent are the spurious activations from extraneous noise? does putting them in sleep mode help?
- what level of recognition are people getting from these devices in 'real' situations?
- what types and what proportion of people find these devices acceptable and what are their reasons for accepting/rejecting them?
- do these devices work for people with speech problems (I'm specifically interested in severe dysarthria)and what level of recognition is possible?
has anyone done any rigorous testing of these devices?
I'd be grateful for your views/experience
Mark
p.s. you may be interested in a research project we are running on speech recognition and dysarthria on http://www.dcs.shef.ac.uk/~pdg/stardust/
Professor Mark Hawley
Barnsley District General Hospital NHS Trust
Gawber Road
Barnsley S75 2EP
Tel: +44 (0) 1226 777726
Secretary: +44 (0) 1226 730000 ext 2689
Fax: +44 (0) 1226 208159
email: [log in to unmask]
-----Original Message-----
From: p=NHS NATIONAL
INT;a=NHS;c=GB;dda:RFC-822=ASSISTECH(a)JISCMAIL.AC.UK;
Sent: 21 November 2002 09:42
To: p=NHS NATIONAL
INT;a=NHS;c=GB;dda:RFC-822=ASSISTECH(a)JISCMAIL.AC.UK;
Subject: Re: voice recognition for "confused of another city
Alan
My experience of voice recognition is that you need to be fairly clued up to
use it. (in fact rather determined really). People feel strange speaking
commands into a microphone - this might be worse for someone with dementia.
Also the error rate is more important. You need to know what to do when it
mis-recognises, in this case dialing the wrong person. I've also found that
with the Pilot, unless you send it to sleep , it will recognise the odd
extraneous noise. It is infuriating for the user to be watching Coronation
Street, the Pilot recognises someone shouting outside and turns over to
bbc2. The user then gets frustrated and their voice tone changes, making
recognition even worse. And if the user's voice tone changes when trying to
wake it up, then they're really stuck!
Having said all this, I ought to say that I still think voice recognition
is good for the right person. The user just needs to be clear about their
error correction strategies.
Does anyone have contradicting experience?
Barry Taylor, Clinical Scientist
Bioengineering Department
Tulley Medical Physics Building, Hull Royal Infirmary
Anlaby Road, Hull HU3 2JZ
Tel 01482 675928, fax 01482 675750
> ----------
> From: Roger Orpwood[SMTP:[log in to unmask]]
> Reply To: A discussion list for Assistive Technology professionals.
> Sent: 21 November 2002 09:32
> To: [log in to unmask]
> Subject: Re: voice recognition for "confused of another city
>
> We have been using picture phones for people with dementia. They look and
> feel just like an ordinary phone but they have largish buttons into which
> you slot a passport photo sized picture. The user just has to pick up the
> phone and touch the picture of the person they wish to ring. From the
> point
> of view of people with dementia they are great because the amount of new
> learning needed is very limited, and they are interacting with a piece of
> equipment that is similar to what they have used for communicating in the
> past. They are only about L22. We have 40 under evaluation as part of an
> EU
> project and they are looking good.
>
> Roger Orpwood
>
> Dr Roger Orpwood
> Bath Institute of Medical Engineering
> University of Bath
> c/o Wolfson Centre
> Royal United Hospital
> BATH BA1 3NG
> tel: +44(0)1225 824103
> fax: +44(0)1225 824111
> [log in to unmask]
>
|