Microsoft's voice recognition technology reduces its error rate

Microsoft is one of the companies that is investing the most in voice recognition technology. They know that there is a lot of future in this type of technology. For this reason, they have been developing theirs for some time. And they have made great strides so far. Today, the company reveals new advances.

Microsoft's speech recognition technology reduces its error rate

Microsoft announces that they have made great improvements in their speech recognition technology. Thanks to these improvements, they place it at the same level of precision as that used among humans. This system has achieved 5.1% in WER (error in the word rate used). A figure identical to that of humans.

Speech recognition

Therefore, the American company has achieved that its voice recognition system makes a level of failure similar to that of humans when communicating. A key moment in the development of this type of technology. And that shows the great advances that Microsoft is achieving in this field so far.

To achieve this, they have used improved verbal and acoustic language models based on neural networks. Combined with two-way memory, acoustic modeling is improved. In this way, recognition based on a prediction of words that can be used based on a communication history is improved.

Therefore, it can be concluded by saying that Microsoft has achieved a voice recognition system as effective as that of humans. In addition, there is surely still room for improvement, so progress and improvements are not yet over. We will see how this system evolves and if in the not too distant future we interact with our computer through voice.