Speech Recognition

[Darren404]

I was wondering if LinuxMCE might have speech recognition in the future. I know a lot of other companies and projects have had a lot of problems with it and the software has been buggy in even the commercial software. Hmm maybe i answered my own question lol. But just wanted to know your thoughts on it.

here is a link I found to some of the different softwares used for it

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

Thanks



 

[tschak909]

I have used Sphinx in a number of private research projects. The current java implementation works pretty well, however, it is VERY difficult to integrate, and I'm probably one of a handful of people on this planet who have experience building language models for it....)

-Thom

[bulek]

I have used Sphinx in a number of private research projects. The current java implementation works pretty well, however, it is VERY difficult to integrate, and I'm probably one of a handful of people on this planet who have experience building language models for it....)

-Thom

Hi,

nice to know. I'm also partly speech recognition researcher... I had another idea of integration - integrate into Asterisk - so any call can be connected to speech recognition - and then, every speech recognition dialog can be made like a phone call (we have it also in Orbiters). One of my students integrated Sphinx with Asterisk in some partial way.... I think this could be shared if someone is willing to push it....

How are your experiences with Sphinx ? I'm really sad, cause open source world cannot offer something decent in this area... There is some pretty simple way to integrate Sphinx2 through recognition of recordings Asterisk could make, but I'm not sure about real performance.

I remember that Pluto did some initial testings with those features, but performance was simply not acceptable at that time....

I plan to put some effort on this - I want to talk to my house .... I want my house also to be intelligent and respond in such manner, event to spoken dialogue.... Some day...

Regards,

Bulek.

[tschak909]

I have used Sphinx in a number of private research projects. The current java implementation works pretty well, however, it is VERY difficult to integrate, and I'm probably one of a handful of people on this planet who have experience building language models for it....)

-Thom

Hi,

nice to know. I'm also partly speech recognition researcher... I had another idea of integration - integrate into Asterisk - so any call can be connected to speech recognition - and then, every speech recognition dialog can be made like a phone call (we have it also in Orbiters). One of my students integrated Sphinx with Asterisk in some partial way.... I think this could be shared if someone is willing to push it....

How are your experiences with Sphinx ? I'm really sad, cause open source world cannot offer something decent in this area... There is some pretty simple way to integrate Sphinx2 through recognition of recordings Asterisk could make, but I'm not sure about real performance.

I remember that Pluto did some initial testings with those features, but performance was simply not acceptable at that time....

I plan to put some effort on this - I want to talk to my house .... I want my house also to be intelligent and respond in such manner, event to spoken dialogue.... Some day...

Regards,

Bulek.

I'd like to try too. It seems that Sphinx is our ONLY option...and it will be ..... an interesting and somewhat arduous experience to integrate this..

There are beginnings of a java based DCE communication library that I saw..... buried deep in Pluto's SVN somewhere.....

hum, investigation is needed.

-Thom

[hodge]

That would be awesome.  Then I could go into a room and do the "Computer lights on" or "Illuminate" from Demolition Man.

[richard.e.morton]

Hi There,

I have previously designed and developed Voice User Interfaces for callcentre applications with Speech Recognition, with technologies such as Nuance 7, 8 and 8.5, Speechworks OSR1 and 2, Philips Speechpearl... and a few others, all of these were eventually bought by Scansoft which renamed themselves Nuance.

So if there is a project in this area I am certainly able to lend a hand with the VUI, it is a specialist area and although not experiences with home automation interfaces I am sure some principles and experience will be relevant.

What we'd really be looking at is multimodal interfaces... speech input and visual feedback either through an orbiter or OSD, MM work is something I have used previously in research work.

Anyway, if this goes anywhere, please feel free to drop me a line and I'll lend a hand...


R

[bulek]

Hi,

I remember that Jim did something with Misterhouse and Asterisk. And it seems pretty painless to try....

http://www.nabble.com/Misterhouse-Asterisk-%2B-Pocketsphinx-HOWTO-td16800602.html

I'm sure that this is not the ultimate speech engine, but it's a start and available to anyone... Will be happy to hear any opinions on this matter

Update: I also found something interesting: playing good old adventure games via speech recognition and synthesis (Sphinx2+Festival) on Asterisk :

http://uc.org/read/ZoIP

Gonna definitely try this one out.... Anyone to join me ?

HTH,

regards,

Bulek.

[acognard]

hi all, this topic is coming back

(sorry for my english)

i am an KNX home automation integrator (starding this new job) and would like to use speach recognition in my house first.
i found this post & read many of you were ok to continue this projecT.

i do not have any experience in programming in a public project, but would be happy to.

i imagine this :
me : computer !
core:yes sir
me:swith on central light in kitchen
core: plse confirm : swith on central light in kitchen
me:yes computer
core:done sir

other scenario :
me:computer
core:yes sir
me:add meeting
core:pls confirm 'add meeting in agenda'
me:yes computer, on 21 th of july 2011, "lunch with simon"
core:done sir

or
core:sir, simon is in bed, may i change temperature to 17°C ?
me:yes computer
core:done sir

(is this realistic or a dream ?)

is asterisk an entry point ? what about ohter systems included in linuxMCE ?
i found http://www.simon-listens.org/ which seems to be very intersting (and sufficient ?)
can you all coming back here and participate to this new challenge ?

Tks in advance.
Arnaud

[tschak909]

Use CMU Sphinx.

-Thom

[lemming86_au]

Saw this project recently: http://julius.sourceforge.jp/en_index.php
Would it be useful?

[tschak909]

It's an engine like Sphinx, except with emphasis on far east linguistic structures.

-Thom

[Ziris]

I, too, am interested in the prospect of having voice recognition and speech synthesis integrated into this project.  From what I understand, CMUsphinx is a backend, much like mythtv has a frontend and a backend.  So all that would need to be done is create a small program that is able to tell the backend to listen for voice, then be able to parse the voice it captured to a command, then be able to pass that command on to whatever program was called in the command.  So, say, "Call Fred" would tell Asterisk to start up, and it would pass the argument "Fred" to it to call Fred automatically.  I was thinking about toying with the notion of integrating this somehow.  Now that it seems that other people are  engaged in this, maybe I could lend a hand?

[tschak909]

feel free to jump in and own it.

-Thom

[bulek]

I, too, am interested in the prospect of having voice recognition and speech synthesis integrated into this project.  From what I understand, CMUsphinx is a backend, much like mythtv has a frontend and a backend.  So all that would need to be done is create a small program that is able to tell the backend to listen for voice, then be able to parse the voice it captured to a command, then be able to pass that command on to whatever program was called in the command.  So, say, "Call Fred" would tell Asterisk to start up, and it would pass the argument "Fred" to it to call Fred automatically.  I was thinking about toying with the notion of integrating this somehow.  Now that it seems that other people are  engaged in this, maybe I could lend a hand?

I guess that noone is working on it actively. If I may suggest, I think going "the Asterisk way" is probably most efficient under LMCE. Tieing speech recognition to Asterisk would be the right thing, since we have embedded phones on MD, VOIP support and Asterisk already in LMCE....

Few starting points :
http://scribblej.com/svn/
http://www.syednetworks.com/asterisk-integration-with-sphinx-voice-recognition-system

HTH,
regards,
Bulek.

[tschak909]

somebody just do it. please. consider it research, and find possibilities.

-Thom