Is Linux a desktop operating system?
Eric S. Johansson
esj at harvee.org
Sun May 29 19:39:15 UTC 2005
Tom Adelstein wrote:
> IBM transferred their voice technology products to Scansoft. Scansoft is
> a commercial software company and if the demand surfaced, I can see them
> porting a lot of software to Linux.
I wouldn't count on it. ScanSoft is the Microsoft of speech
recognition. They are acquiring lots of companies, eliminating choice,
and maximizing profits. They have created a monopoly in the desktop
speech-recognition arena. They're working towards creating a monopoly
in the telephony (i.e. small vocabulary, small grammar)
speech-recognition Arena.
It appears the internal attitude towards Linux ports was that to do so
would cause them to lose all of their intellectual property protections.
Attempts to educate them otherwise were not well received. As far as
I know, this attitude persists to this day.
they're not longer interested in working with a handicapped community.
They have no visible outreach program, they are not fixing critical bugs
that impact the handicapped community (i.e. natural text, sapi 4 grammer
failures). Since NaturallySpeaking 6, functionality has gone down,
speed has decreased, and the bug list keeps growing.
A few years ago, NaturallySpeaking worked relatively well with a fair
number of applications. Today, NaturallySpeaking well with a smaller
number of applications (i.e. Microsoft Word, Explorer, and Outlook).
ScanSoft has added activation codes so you can only install the software
on a finite number of machines and a finite number of times. The
current suspicion is this is the way they will enforce mandatory
upgrades to maximize their revenue from the people that can least afford it.
Needless to say, I am not happy about the state of affairs because it's
the only way I can use computers to any great extent. You have no idea
how enraging/discouraging/infuriating it is to be this dependent and
then to be exploited because of it.
I've been living in this world ever since I was injured in 1994. I've
seen it go from discrete recognition through continuous and I know they
can do better than they are doing now. But open source recognition
systems will not be something you do on the cheap. Or as result of the
few grad students knocking out a prototype in a semester. It's going to
be hard work for a long time. There'll be numerous patents to bypass
and usability challenges. I don't say can't be done but it's something
you will need millions for.
in the short term, I am looking for alternatives that let us run
NaturallySpeaking on Linux.
--- eric
More information about the ubuntu-users
mailing list