BrainStream: V2V/ITS and Voice Recognition

[bkdelong]

Well, apparently I am incredibly far behind with regard to vehicle-to-vehicle communications and intelligent transportation systems. So it's not like I'm the only one thinking up these ideas.

However, it did get me thinking about using Voice Recognition with such systems. I mean, we all abhor folks talking on cell phones while driving - making use of ITS while in motion would be a nightmare.

But if you think of all the systems one could potentially use in the future, with voice recognition - one must wonder how long it would take to continually "train" these systems to understand what you're saying. That's where my idea for Voice Recognition Profiles (VRP) comes in - still looking to see who else has done it.

So when I load up a voice recognition program, I am told to read several lines or paragraphs of text so it can match the text content with my voice. For every program I try, I have to retrain it all over again. In theory, if I move from my computer to my car and try to activate my GPS system by voice, it needs to be trained. If I go to an ATM or drive-thru where one can automatically order by voice, I need to spend several minutes correcting the system until I'm connected with a human operator because the damn thing can't understand me.

Why not create a standard profile for voice recognition that all voice-recognition applications can use? That way, when I come to a new system I need to "train", I just type in my SSN or some other UID which tells the system to pull my VRP (Voice Recognition Profile), out of a centralized directory service, allowing me to immediately use the system with a peak understanding of my voice.

In theory, each time I access a new service using my VRP, whatever actions I take and corrections I make in the process, would be noted in my profile and sent back to the directory service for the next time I access a service - a live, constantly-growing, learning profile.

The futurist in me sees the next step to that being appending a subvocalization profile which would translate the subvocalization signals directly to something that could be used to access various devices around an individual, perhaps an enhanced version of Bluetooth.

Anyone heard of such efforts to develop such a voice profile?

Comments

Intelligent Transportation Systems (ITS) and Stds.

[ocean-portal.livejournal.com]

I'm not sure how much your have read on ITS but here is a USDOT site to start at. It links to various Govt. labs doing research for various transportation modes.

http://www.its.dot.gov/index.htm

Re: Intelligent Transportation Systems (ITS) and Stds.

[bkdelong.livejournal.com]

Thanks - yeah, I did some pretty quick research on V2V and ITS with regard to the DSRC standards. It looks like if the data generated from such activities are going to be open and easy to access then the Open Source community will have to take some degree of ownership and involvement.

Dragon Preternaturally Speaking

[(Anonymous)]

Well, sure; David Pogue has a nice story about having to retrain as his voice and RSI got progressively worse (culminating nicely in a TMJ/RSI/hairball release event...something about getting up to adjust the shades.)
Naturally you want to have to retrain words as little as possible; however as I recall Dragon and IBM's efforts to go to ISO on this case did not land on terra firma (though they each claim standards, and IBM has orthogonal note and word recognition tools here and there. Couldn't find the basis myself....

A corpus isn't particularly great, either; without a bunch of contextual info on what you were speaking on and where, you hardly know what to think of the way you sounded (or should I keep this to myself?) So I think that it's a matter of what I've decided to slur and why I might be slurring it and throat-singing in the background amain that the computer needs to be better at ignoring or mirroring or something. Yes I'm twiddling a full markov graph at a time, box! Kep up!