2 ms·
From a pure Speech Recognition POV, I will believe Google Goggles is possible when they somehow solved the major problems in Speech Recognition. SR tech has bee
by swang 15y ago
From a pure Speech Recognition POV, I will believe Google Goggles is possible when they somehow solved the major problems in Speech Recognition. SR tech has been around forever, but only in the last few years have companies been able to process it on the server side and then zap it over a wireless connection to get somewhat OK processing times.
But no one has yet to solve the two problems plaguing SR: First, putting a whole SR engine on a machine let alone on a pair of embedded chips on a piece of eyewear for instant SR. Even with miniaturization and speed/power improvements I just don't see it happening mainly because chip makers have decided to just add more cores. I may be wrong but it seems difficult to parallelize this with a large corpus.
Obviously the solution is to go the other way and have super fast guaranteed wireless connection at all times. This will still produce lag but probably acceptable levels for most people.
The second problem is one that most people encounter with SR which is when the bloody thing doesn't translate what you said. Either because of an accent or because you're saying something unique or difficult to parse. I tried asking directions to a Mexican place in OC and the results were: Q Cortes, cute protcullis, cute Portos, Q Portales. Can you figure out what I was trying to say? That is the inherent difficulty of the "last mile" problems that SR has and I haven't read or heard anything about anyone solving this anytime soon. SR tech will always be just "good enough" where they can show it off at a meeting/conference but never good enough to be like Star Trek (which unfortunately has set a incredibly huge bar for people's expectations).
Even though the Nokia video is less ambitious they seem to have decided to not include any talking in their future tech videos which seems like a smart idea to me.
- grogs 15y agoMaybe the solution to SR is more/better learning. The computer could ask for confirmation/clarification, and gradually learn your speach patterns over time - lessening the need for constant clarification. This sounds achievable. People in Scotland should be able to talk to Siri, the learning phase should just be a little longer.
- JohnLBevan 15y agoI guess if you have a central SR processor which everyone's connecting with (as per proposed solution for issue 1) then you also have a huge amount of incoming data from all of your users. That means your system can evolve by taking feedback on errors its made, and can rapidly improve based on this new information, building larger dictionaries, taking info such as location data to improve context awareness, and using the vast volume of samples to improve SR where unusual accents are involved.
- stcredzero 15y agoFrom a pure Speech Recognition POV, I will believe Google Goggles is possible when they somehow solved the major problems in Speech Recognition. I think one could take the form factor and have an awesome product without speech recognition. The potential for AR games alone is compelling. Real time translation of foreign language signs would be useful. Tourist guidebooks would be great. HUD for maps/gps alone could be a viable market.