Google virtual brain tech set to be used in speech recognition
Google’s virtual brain technology was shown off earlier this year, when engineers revealed it was capable of recognising people’s faces and even cats just by watching YouTube clips. And now the tech is being rolled out into actual Google products, with speech recognition services the first to get the treatment.
Called a ‘neural network’, this new Google tech aims to work much like a human brain, with cells communicating with each other independently, without human input. In terms of speech recognition, this tech can now recognise more words, more accurately, meaning users won’t face as many nasty niggles when using Google’s speech services.
"We got between 20 and 25 percent improvement in terms of words that are wrong," said Google’s neural network guru Vincent Vanhoucke. "That means that many more people will have a perfect experience without errors."
Word is this tech will also be able to help with such futuristic projects as Google Project Glass and the company’s self-driving cars. For now, the neural network-backed speech recognition tool will only work with US English, but expect to hear plenty more in the coming months.