Embedded Speech Might Soon be a Reality


Imagine you have bought a new toaster, but it is unlike any other toaster you may have ever seen. When you take it out from its box and power it up, the toaster asks for your name. When you drop bread into the toaster, it asks you how you would like the bread to be toasted.

When the toast is ready, it asks you how it tasted and you give it some feedback. The toaster remembers the feedback and the next time you put bread into the toaster, it toasts the bread just as you want. If you change the bread, the toaster recognizes even that and again asks for your preference. Over time, the toaster records your preferences for wheat bread, white bread and so on. This technology is called embedded speech.

The good news is, the technique will be a reality soon. It may take some time to see it in our kitchens and homes, but it will ‘arrive’. There is already a prototype in the market called MOVI. The makers are trying to find investors for the product. MOVI is able to recognize hundreds of sentences, right from simple commands like, ‘Light on’ to ‘Turn the ceiling fan on’. Another interesting thing is, MOVI is speaker-independent, which means it will respond to the sentences no matter who speaks them.