If you have ever used a speech recognition device, you must have experienced a high rate of errors in the software. Fortunately, researchers at Microsoft Corporation (NASDAQ:MSFT) may just have outdone themselves, as the company has revealed details of a new artificial intelligence speech recognition system which can transcribe conversations as well as a human being can. Head of AI speech research, Xuedong Huang, has stated that the company’s newly developed AI speech recognition system is a historical achievement for the Redmond based tech giant.
Interestingly, Microsoft has confirmed that its new system has a error rate of just 5.9%, which happens to be very close to what you can expect from professional transcriptionists. The company has revealed that its new speech recognition system is programmed to use various neural language models to piece different words together in a group which allows the system to efficiently generalize the data that it is fed with. Also, the software giant plans to incorporate its new artificial intelligence speech recognition software with Cortana, which is the company’s personal digital assistant for Windows and Xbox One users.
Even though Microsoft’s new AI speech recognition software represents a major technological breakthrough for the tech world, the tech giant cannot afford to rest on its laurels, as every major tech company is reportedly working on perfecting its AI systems. Microsoft has only managed to complete half the task, as the company still needs to teach its AI system how to effectively act upon human speech rather than just understand what it is told. Hence, Microsoft and the entire tech world still have a long way to go.
Microsoft is set to kick-off its next major event by the end of October, so there is a high chance the company will make further announcements regarding its upcoming speech recognition software. Stay tuned for more, as we keep you posted with the latest news surrounding Microsoft’s latest projects.