Microsoft AI reaches 'human-level' speech understanding

Staff astounded by speed of innovation on AI speech recognition

Microsoft AI researchers have claimed that the latest generation of speech recognition technology being developed in the company's labs is on a par with that of a human being.

"Microsoft has made a major breakthrough in speech recognition, creating a technology that recognises the words in a conversation as well as a person does," the firm said in a blog post.

The team showed in a paper published yesterday entitled Achieving Human Parity in Conversational Speech Recognition that it had reached a word error rate of just 5.9 per cent, which is equivalent to an average human being.

The paper explained that this is something of a holy grail in machine learning and artificial intelligence, as well as a huge boost to those with disabilities who perhaps can't use a keyboard.

A number of other companies, such as Google (of course) and Nuance, have demonstrated big leaps and bounds in the use of deep learning to bolster the ability for machines to interpret human speech using a knowledge of context and accent to ensure the right result.

Nowhere is that more obvious than the game-changing Amazon Echo, which has brought speech-controlled computing to the home for £50.

"We've reached human parity," said Xuedong Huang, Microsoft's chief speech scientist. "This is an historic achievement."

Harry Shum, executive vice president of the Microsoft Artificial Intelligence and Research group, added: "Even five years ago, I wouldn't have thought we could have achieved this. I just wouldn't have thought it would be possible."

Reaching human level parity is good, but Microsoft admitted that it is not perfect, explaining that, like humans, the system still makes mistakes.

However, unlike humans it may be possible for the team to improve the speech recognition to the point where no mistakes are made.

But perhaps the real challenge will come with understanding double meanings and word plays, something that may be beyond the abilities of computers and remain the preserve of humanity. Perhaps.