Accessibility and Coolness - Voice Recognition in Linux

[cbhworld]

A little while ago, Jono said that the open source community couldn't do ground breaking research. In a separate episode, Ade said that he'd be talking to his Linux computer - rather than using the keyboard or typing, in about 50 years from now.

What can I make these have in common? Voice recognition software. Now - voice recognition is not ground breaking, and there are already some decent open source projects (or pretty much one...) that do this - like the Sphinx projects, in all their glorious incarnations. And loads of closed source commercial products too.

Anyway, does anyone here have any experience - knowledge or - well - experience or knowledge of voice recognition for Linux? (Or emulating Windows programs...) The reason I ask is because my father is very interested in Linux but is unable to use it because he relies on voice recognition due to his combined physical impairments and dyslexia. Despite not really being in the 'communities' that surround Linux and open source - my dad has tried to get involved and has created a discussion thread on his website: http://www.hi2u.org/discussions/voice4linux.htm about this problem, and asking for any information.

If anyone could add anything useful to that, it'd help - if not, just put it here and I'll do it.

Anyway, thanks for reading this post - I'm hoping that someone out there might be able to bring Linux closer to my dad.

Chris Hayes / CBHworld

P.S. I think I saw something about MrBen having been involved in the field of voice recognition? I could be wrong...

[firefly2442]

Sorry, I don't know of any voice to text software out there. I do know of "festival", which does text to speech. But I don't know how helpful that is...

I hope you are able to find something that works well for your father.

edit: google search found these:

http://xvoice.sourceforge.net/

http://linux-sound.org/speech.html

http://www.faqs.org/docs/Linux-HOWTO/Sp ... HOWTO.html

[neuro]

Edit: I suck.

[cbhworld]

Thanks for looking - but that's reminded me of another good point - about 4 years ago, voice recognition on Linux was looking really positive - IBM was releasing an edition of Via Voice for Linux and even open sourced parts of it so that programs (such as xvoice) could use it's engine. Since then our corporate gurdians of the open source world, IBM, pulled out and it's difficult to even get hold of what they did release now.

Indeed, 3 or 4 years ago it all looked so promising. You'll notice that the HOWTO you posted a link to, firefly2442, is drasticaly out of date - it's actually a project of mine to find some time and update it myself.

Thanks for having a look though - I'm suprised how few people seem to be interested in voice recognition technology, I mean - as far as nerd points, it's cool, and (without any experience or real knowledge about it) the sphinx projects looks pretty good, and I've talked to some people who have tried them, and they say that they work reasonably well - they just aren't for real people yet - I bet all someone needs to do is create a GUI frontend - so normal users can at least try the technology - would generate enough interest in the project to polish it off enough to be useful....


Chris Hayes / CBHworld


Any thoughts? And I'm still really interested in any information or experience in this area from absolutly anyone.

[milesteg]

Not that I can offer any more positive advice, but along with linux's existing accessibility features (such as screen zooming for the visually impaired and so on), good voice recognition software would greatly improve the os's usefulness for those with disabilities...

I do wonder sometimes why so many programmers direct their efforts at developing music players/torrent feeders and so on (when we have enough of them already, and none of them actually completed/bug free) when there is no voice recognition software, no GUI for PGP software and other things that would be really useful but which are simply absent from linux distros

[neuro]

[snip] no GUI for PGP software and other things that would be really useful but which are simply absent from linux distros

What's wrong with seahorse?

[Allix]

Since then our corporate gurdians of the open source world, IBM, pulled out and it's difficult to even get hold of what they did release now.

IBM have adapted to open source, they are not the guardian in anyway to opensource.
In the beginning there even wanted to write there own license which was very much not open source in the freedom sense.Bruce Perens did substantial work in help the rewriting of there license so it became acceptable.

Thanks for having a look though - I'm suprised how few people seem to be interested in voice recognition technology, I mean - as far as nerd points, it's cool, and (without any experience or real knowledge about it) the sphinx projects looks pretty good, and I've talked to some people who have tried them, and they say that they work reasonably well - they just aren't for real people yet - I bet all someone needs to do is create a GUI frontend - so normal users can at least try the technology - would generate enough interest in the project to polish it off enough to be useful....

It really comes down to resources , its obvious that the developers who are interested in this technology are not interested in open source at least from what you are saying.
Its ashame because it an area that open source would benefit.
On a side not , where i study, robotics is very big and the university does open source the hardware and software, at leasts thats from the posters i see.
Apart from a learning environment Open source does not seem to attract new technology.

[cbhworld]

Allix wrote:

cbhworld wrote:

Since then our corporate gurdians of the open source world, IBM, pulled out and it's difficult to even get hold of what they did release now.

IBM have adapted to open source, they are not the guardian in anyway to opensource.

I didn't know that - but you can't deny that quite a few people (blame the media...?) seem to think that they really are an open source company, more like what Sun is.

I don't think that the current Sphinx projects are too far away from being useful if someone would wrap a nice GUI around one... Isn't the Ultimate edition (and perhaps some of the lower versions) coming complete with voice recognition software now? Much like the voice recognition software that has been bundled with at least some of the modern Microsoft Office packages...

[milesteg]

sorry, dragging off topic here - with seahorse you can't create encrypted volumes though can you? That's the functionality I was really looking for - although I see that it offers a gui to manage your keys so I may have a play with it anyways

[neuro]

sorry, dragging off topic here - with seahorse you can't create encrypted volumes though can you?

You didn't mention that tho, did you?

[cbhworld]

Well, *gently nudging back to the topic* it seems that this HOWTO ( http://taint.org/wk/ViaVoiceModernLinux ) details ways of getting the old IBM ViaVoice for Linux working on a modern distro - and where to get - some of - the bits from. I'll have a go at getting this working shortly. Whether it's legal or not I wouldn't know...

Also, it appears that IBM still sell their voice technology for cross-platform development... http://www-306.ibm.com/software/pervasi ... iplatform/

It's nice to see that IBM's interest in Linux is to serve companies rather than consumers.

Chris Hayes / CBHworld

[Allix]

It's nice to see that IBM's interest in Linux is to serve companies rather than consumers.

how is that nice to consumers ?

[cbhworld]

Oh, I'm actually corporate whore underneath my loving vegetarian visage, Allix. I care not for the consumers really.


Chris Hayes / CBHworld

[Allix]

Oh, I'm actually corporate whore underneath my loving vegetarian visage, Allix. I care not for the consumers really.


Chris Hayes / CBHworld

hehe