Roy Schestowitz wrote:
> __/ [ Geico Caveman ] on Wednesday 16 August 2006 16:21 \__
>
>> I have been a Linux user for years, but there is one useful functionality
>> that has mostly been missing on Linux - voice recognition and dictation.
>> Ever since IBM discontinued ViaVoice for Linux (I still have my old CD
>> set in fond hopes that I might be able to use it - but each glibc and
>> java (I do not understand this obsession some commercial vendors have
>> with java - makes everything slow and buggy) upgrade makes that more and
>> more unlikely),
>>
>> 2001 was a heady time for this - ViaVoice and XVoice. I remember
>> dictating some of my longer reports into emacs at the time. But right
>> now, there appears to be absolutely nothing. Google search just yields a
>> shitload of outdated pages about ViaVoice. CMU Sphinx is not mature and
>> there are some rumors that it has been abandoned. I tried a version of it
>> a few months ago and it is kid stuff (putting it mildly) compared to what
>> ViaVoice was.
>>
>> I do not much care for games, video editing etc. (even though decent ones
>> in either exist for Linux). But voice recognition is a very useful
>> function to have, especially for busy people who sometimes need to type
>> long documents. I do remember that there were some rumors a few months
>> ago that the new version of KDE (I am *NOT* talking about kttsd) was
>> going to include this functionality, but I have not read anything since.
>> Maybe it was just limited stuff for hotkeys
>> (http://developer.kde.org/summerofcode/speech.html). Is there any hope
>> for voice recognition / dictation software (preferably open source) for
>> Linux any time soon ? And no, Wizzscribe is NOT it - it is something for
>> developers and even that costs more than $2K. Out of reach for a normal
>> user.
>>
>> There is an older report about voice recognition with Qt :
>>
>> http://linuxpr.com/releases/3255.html
>>
>> But since IBM bailed out, and there has been no talk since about Qt
>> (checked Trolltech's website), I do not imagine anything came of it.
>
> It all changed very recently.
>
> http://linux.sys-con.com/read/249714.htm
> Wizzard Software Releases WizzScribe SI For Linux
>
> Revolutionizes Server-Based Speech Recognition By
> Eliminating Training
>
> http://biz.yahoo.com/iw/060807/0149614.html
> Speech Recognition Software Adoption Powers Voice
> Applications Around the World
>
> http://biz.yahoo.com/iw/060807/0149620.html
> Web-Based Platform Revolutionizes Voice Business
>
>
> There are more projects, some of which are free. You can find the links
> rather easily...
>
> Meanwhile, the competitor is making 'baby steps'.
>
> http://video.google.com/videoplay?docid=-1123221217782777472
>
> Best wishes,
>
> Roy
>
You missed this :
"And no, Wizzscribe is NOT it - it is something for developers and even that
costs more than $2K. Out of reach for a normal user."
Can you point me to one open source / even commercially supported today
piece of software, which will allow me :
1. To train the engine.
2. Dictate my technical reports into emacs.
?
I checked up on wizzscribe (full of hope) the hour you posted the news 3-4
weeks ago. It is NOT it. It is for developers/big businesses and it costs
about $2K. Totally out of reach for a single user like me and compared to
its possible competitors on other platforms - grossly overpriced as well as
being a hammer for cracking an egg. My IBM ViaVoice box cost me $70 odd in
2001, IIRC. Its not even completely clear what Wizzscribe does (I am
looking for a plain dictation software with speech to text capabilities.
Period.).
While I appreciate your promotion of Linux even more than I do, it is
equally important to realize where it does not (yet) offer a solution.
Dictation with speech-to-text is one of those few areas that really matter
where there are no solutions for Linux (open source or otherwise) for an
ordinary user.
|
|