Microsoft has announced some major improvements to its Bing Voice Search for Windows Phone devices. According to an announcement on its blog, it has developed a new technology on which the latest iteration of Voice Search is based upon. This technology has been dubbed as Deep Neural Networks by the software giant. DNNs have helped to greatly improve the speed at which results are returned to the Windows Phone devices and have also allowed to reduce the word error rate by more than 15 per cent.
These changes make the voice search not only more accurate, but also bring it closer to much more easier reproduction of speech recognition, similar to what happens in the human brain which is also made up of neurons. This biologically based model has allowed it to cut the response time into half and allowed it to work in more natural environments with more ambient noise.
According to another post on Microsoft Research website, this technology even allows learning across multiple languages, so that data from one language can significantly affect the learning patterns in other languages.
Dong Yu, senior researcher at Microsoft Research Asia said, “Before our result, deep learning was only tested on small tasks and did not attract wide attention. I believe this is just the first step in advancing the state of the art. Many difficult problems may be attacked under this framework, which might lead to even greater advances.”
These latest updates to the Bing search should have already been delivered to most Windows Phone users via an update.
