New milestone in teaching machines to understand human speech - Asia News Center

The quest to teach machines to understand human conversations has taken another big step forward with researchers achieving a new level of speech recognition for technology. Xuedong Huang, who leads Microsoft's Speech and Language Group, announced the new milestone in a blog post. Last year his team made headlines when it reached human parity on the "Switchboard conversational speech recognition task", meaning it had created technology that could recognize words in spoken conversation as well as professional human transcribers. Switchboard is a corpus of recorded telephone conversations that the speech research community has used for more than 20 years to benchmark speech recognition systems. The task involves the technology transcribing conversations between strangers who discuss various topics, such as sports and politics.