GNOME.org
 

Engine on PC - on phone would be great! Re: applicable Re: Voice recognition for dashboard

Voicemail was just an example I gave. I am afraid most of the Linux community does not use voice/speech recognition for the simple reason that ther are no good engines out there yet. But because I think more than I could possibly write, I would really like to have a dictation engine for myself. If anyone knows of a free engine or has a spare copy ;-) please let me know.
Coming back to my thoughts. I am from India and presently doing a PhD in Glasgow but I use Yahoo Messenger, MSN Messenger and MS-Netmeeting to talk to my folk back home. I am sure that this kind of using speech to communicate using the computer (SIP phones and IP Telephony) will get to the Linux world too. Al, you mentioned GAIM. Never got that of GNOME-Meeting to work. When I talk, I would want the ASR Engine to identify clues (whatever it can to whatever accuracy it can) and these could be useful. Very useful if my understanding of Dashboard is correct. As per your example, if the ASR Engine running on my PC/PDA can identify the date, this "clue" could look-up my calendar and tell me the event at that time or even search my bookmarks or task-lists. What one can do with clues is really upto the backend applications. The front-end applications if they use voice could use ASR to some good effect. So, the engine need be on the PC which is doing all the indexing.
Yes. SIP headers are good too and very interesting indeed. If we can get real-world hardware like phones, stereos, printers, MP3 players, microwaves etc. give out clues which could be captured by our computing systems by Bluetooth or IR or RFID or whatever, that would be great. Again, these might be seen as front-ends broadcasting clues. Once the computing system has the clues, it is all left to the backend applications not only to use the clues but also to use the context (which device) to do some intelligent processing. 

----
Regards,
Srikant
(http://sriks6711.blogspot.com)
(http://www.dcs.gla.ac.uk/~sriks)
"Arguments are a good way to achieve progress. Accidents and dreams are another ways" 
-=-Sriksism's powered by TagZilla-=-

Al Baker wrote:
Where is the speech recognition application though? Obviously, it's probably not on the voice mail server. It would have to be on whatever application you were running on your linux desktop that let you listen your voicemails with, like one of those linux SIP phones (linphone or that modification to GAIM).
In either case, if there's a phone application that's sending clue packets there's probably better [and easier] things to get. For example, a phone could send clue packets based on the call itself (name of the caller, time of call, duration of call, if the call was answered or not, etc). All of this information is available via standard headers in the SIP protocol.
Getting back to ASR, I guess you could have some ASR filtering your voice stream in the phone and it could add those extra clues for those dates, times, and locations -- but it wouldn't give any other context to that clue. For example, if you dialed into a voice mail system you probably already know who left the voice mail and when it was left. If you're listening to a voice mail message, like "Hey Srikant, the concert is on the 19th at 8 o'clock", I'm guessing that the ASR is going to be able to recognize "19th" and "8 o'clock", so programmatically you have whatever information you get from the voice mail server about the message (who left it and when) and some of this data -- but it wouldn't have your name or that it was a concert [unless concert appears in the locations grammar].
At the moment, I don't think there's any linux voice mail application - either stand alone or as a feature in linphone/gaim's SIP version). So, if you wanted your voice mail you're probably dialing an external number to some external mailbox and getting your voicemails through a prompt system. The reason I bring this up is because there'd be no saving of voice mail messages on the user's PC. So, there'd be no way to easily launch something to hear the voicemail once you found the information in dashboard.
The more I think about it, getting SIP support into GAIM and then getting more clues from GAIM would be a really great thing. 

Al

On Fri, 2004-09-17 at 07:31 +0100, Srikant Jakilinki wrote:
I would say that the dates, times and locations - are always important as clues. In technical terms, they are called "named entities". The IBM package would become uber-cool if we could identify names of people as well. And it would be nice to integrate all this with telephony. Like say, extracting information out from our voicemail boxes. I also think that we should have a page on Dashboard to collect all these links. They are just embedded somewhere deep within the mailing list. On Friday September 17th 2004, Al Baker wrote: > What IBM is releasing sounds like it will have a limited grammar and > grammar detection support in it. The article mentions it would > primarily recognize speech for dates, times, and locations -- e.g. for > reservation systems. Probably not very applicable to dashboard until it > could do the things you describe below. > Just my 2 cents. > Al > On Mon, 2004-09-13 at 22:54 +0200, Nils Erik Svang� wrote: > > Hi! > > I read on slashdot that IBM is releasing a voicerecognition packet as > > opensource. > > It would be very cool if dashboard could create cluepackets from the > > speechrecognition software and dig up info on what whatever you are > > talking about. > > I would be cool for a radiostation, they could get cool trivia on > > whatever the callers or dj is talking about. > > And it would be cool if you just could pipe your favourite > > radiostation throught dashboard and get relevant info as you listen! > > ;) > > > > =) > > /nisse > > _______________________________________________ > > Dashboard-hackers mailing list > > Dashboard-hackers gnome org <mailto:Dashboard-hackers gnome org> > > http://mail.gnome.org/mailman/listinfo/dashboard-hackers > _______________________________________________ > Dashboard-hackers mailing list > Dashboard-hackers gnome org <mailto:Dashboard-hackers gnome org> > http://mail.gnome.org/mailman/listinfo/dashboard-hackers 

--
Joey: Well so, how come you blew me off? Y'know? How come you were with him? Kate: I don't know! I just, just do this! I-I always have to pick the... like the smartest guy, or-or the most talented guy. Why can't I just pick someone like you? Joey: Thanks. Excerpt from the TV Show "Friends" _______________________________________________ Dashboard-hackers mailing list Dashboard-hackers gnome org <mailto:Dashboard-hackers gnome org> http://mail.gnome.org/mailman/listinfo/dashboard-hackers
[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]