Microsoft’s Voice Recognition Tech Reaches ‘Human Parity’
Major improvements are on the way to Microsoft’s voice assistant software Cortana, courtesy of a breakthrough by the company’s researchers and engineers.
The team at Microsoft Artificial Intelligence and Research released a report detailing the results of their new speech recognition system earlier this week, hitting a word error rate of 5.9%.
This puts the software at the same level of proficiency as professional transcriptionists.
“This is an historic achievement,” said Xuedong Huang, the company’s chief speech scientist.
“We’ve reached human parity,” he said.
The company says the milestone will have major implications for both consumer and business products including consumer entertainment devices like the Xbox as well as accessibility tools such as their instant speech-to-text transcription and personal digital assistants such as Cortana.
The team says their next focus is to teach computers not just to transcribe the sounds coming out of a users mouth but interpret and understand them better.