by Srdjan Misic | March 05, 2014
Speech recognition has been capturing the imagination of SF writers and engineers since the dawn of the computer. Throughout the years, transistors got smaller, code more sophisticated, and today voice activated home automation is the subject of many DIY projects for enthusiasts all around the world. Usually it involves some kind of workaround with a Raspberry Pi and/or cloud based software available on your smartphone. But there must be a better way.
We've been thinking about it.
There's something emotionally rewarding in controlling things with your voice. It's similar to having an obedient dog. You train your dog to listen to commands spoken in your own language, you don't try to bark at it. The dog doesn't care, it doesn't understand what your words mean, it just correlates an arbitrary sound with a specific action. You of course know this, but you can't help but feel good about having that control, having your pet adapt to your preferred means of communication. Whereas when you push a button or press a key, you go out of your way to make it simpler for the machine.
So it's a tempting thought, but 9 out of 10 people usually discard it, because frankly, it's not worth tinkering that much to hack a solution for your own home, when a simple switch has been doing the job fine for the last hundred years or so...
But, what if we had a click board for it? We'd make it dead simple to program and calibrate to recognize your voice, whatever your language, local dialect, or choice of pop culture reference is. Make it able to understand and execute over 200 commands. Now let's give it a few GPIOs to make it work as a standalone device too.
Use your voice to turn relays on or off, trigger the alarm, turn on lights, change TV channels, open doors with safe-words etc. All on your own terms too: it'll be speaker dependent, reacting only to your voice, not like some tail spinning tongue wagging dog happy to please any old burglar that comes by.
Well, we're working on exactly that. We call it the SpeakUp. We're taking speech recognition and making it simple. It'll be out soon, it's just that we couldn't hold our enthusiasm so this is an early announcement. Stay tuned for specs and details coming out soon.
by Srdjan Misic | October 20, 2016
The software roadmap has been updated with new info for our most popular compilers.[more info]
by Srdjan Misic | October 14, 2016
Today 24 students and three professors from the Faculty of Technical Sciences in Novi Sad visited us.[more info]