Imagine you can control everything just by saying something.
Speech recognition was a great challenge to humanity for the last 50 years, and still is. And we are about to solve that.
Instead of a pure statistical approach, we decided to study the process of human producing and recognizing sounds.
Our first results allowed us to make classifiers for vowel sounds. Now that we have developed the core of our technology and are knowledgeable how to develop classifiers for other sounds, we strive to achieve more and more sounds every day. Our goal is to create classifiers for every sound a human can make which is about 200 sounds. This way we can support all languages and accents.
After we can classify sound with a good certainty, we are going to add statistical search methods to finalize the recognition process.
We created a functioning Android library for sound recognition. Stay tuned, soon we will have a test game app release on the app store.