Microsoft has claimed that Bing voice recognition for Windows Phone has been overhauled so as to perk up the speed as well as accuracy for the speech to text and voice search features. A post on the search engine’s official blog said that they have joined forces with MSR (Microsoft Research) to tackle the shortcomings which have been faced by the tool so far.
So this is how it works. The efforts of both the teams has led to the inclusion of Deep Neural Networks (DNN), a technology that is said to reflect the attributes of neurons in the brain. To put it simply, the tool will now recognize patterns just like biological systems do.
According to the conglomerate, this has boosted the search and speech-to-text input speed by two folds and the word-error rate has been improved by 15 percent. The updates for the Bing voice recognition feature that have been made available in the past few weeks for Windows Phone owners have also led to proffering relatively accurate content even in a noisy environment.
It is this technology with which Rick Rashid, the company’s chief research officer, was able to host an event in China without worrying about the language barrier. Thanks to DNN, the translator app not only converted English into Mandarin in real time with no grammatical errors, but it also rendered his speech in his own voice. The high point of this software is that it whets itself across languages which means that content from one dialect results into increased accuracy in another.
As of now, Bing voice recognition has been bumped up only for those employing devices powered by Windows Phone in the US and the software maker has made no statement concerning a plausible roll out for people residing in other regions across the globe. But of course we are expecting the tech giant to introduce the advanced features in other regions sooner rather than later.