Researchers over at Microsoft have just divulged details pertinent to their new speech recognition technology that they claim transcribes conversations just as good as humans have. Xuedong Huang, Microsoft?s chief speech scientist stated, ?We have reached human parity. This is an achievement.?
The new program?s error rate reportedly is 5.9%, which according to Microsoft officials is about equal to professional transcriptionists. The system employs the use of neural language models that clusters similar words together for generic efficiency. There are also plans to bring this technology over in Cortana, voice assistant for Xbox One and Windows.
This breakthrough albeit innovative in Microsoft? part, still needs a little buffing to be brought to a wider range of mass consumption.
Comments
Comments are closed.