Microsoft’s New Achievement In Speech Recognition System

Microsoft has taken a huge step forward by making another significant progress in artificial intelligence.


Speech recognition applications are nowadays very sharp but can’t hear the language as precisely as research workers who have created it. Microsoft has overtaken that difficulty and the improvement which was explained in a paper released Monday indicates the most sophisticated speech recognition applications to date, based on Microsoft, who’d also set the previous record for speech recognition.

When they first examined it, they got 6.3. Now, after one month, it’s approximately 5.9%. Keeping the products of consumer apart, it indicates a turning point for research of AI also.

The landmark has far-reaching consequences. On a level of practicality, it means that the products of Microsoft could soon be a great deal better at comprehending people.

The applications itself depend on profound neural networks. Technology that interprets information in a way similar to the way the human brain functions along with specialized GPUs (graphics processing units) that enable the applications to learn at speeds which were not formerly possible. Having the ability to transcribe human language correctly is a huge step forward though that is a target that is considerably farther away.

The researchers termed the Xbox and personal assistant program of Microsoft, Cortana as two products that could immediately reap the benefits of the research. Availability applications, like transcription services that are immediate, could also receive the benefits of the progress.

It could also readily be integrated into productivity applications of Microsoft like Office which is the better dictation attribute of the world. That would be with near-human levels of precision.

In a statement, the Speech and Dialog research group of Microsoft stated that the next stage would be to help develop the software that cannot simply transcribe language that is human but comprehend it at the same time.

The team from Research and Microsoft Artificial Intelligence reported a speech recognition system that produces the same or fewer mistakes than professional transcriptionists.