Microsoft published a research paper on 17 October wherein a team of researchers and engineers of the company’s Artificial Intelligence and Research reported a speech recognition system that makes the same or fewer errors than humans who professionally transcript files do.
The researchers reported a word error rate (WER) of 5.9 per cent. With this accomplishment, the company says the computers can now recognize the words in a conversation as well as a person would.
“Even five years ago, I wouldn’t have thought we could have achieved this. I just wouldn’t have thought it would be possible,” said Harry Shum, the executive vice president who heads the Microsoft Artificial Intelligence and Research group.
This technology will have broad implications for consumer and business products such as Xbox, instant speech-to-text transcription and Cortana among other products.
The Microsoft team made use of Microsoft’s Computational Network Toolkit, a homegrown system for deep learning to reach this milestone. The company said, ‘CNTK’s ability to quickly process deep learning algorithms across multiple computers running a specialized chip called a graphics processing unit vastly improved the speed at which they were able to do their research and, ultimately, reach human parity.’
The company is now working on ways to make sure that the speech recognition works well in real-life settings that include additional external factors such as background noise. They will also focus on ways to help this technology assign names to individual speakers when multiple people are talking.
