Microsoft's technology works with neural networks to improve speech recognition

neural

Using a new method that mimics the human brain, Microsoft researchers claim to have doubled the speed at which the speech is recognized and the results returned to the user. To achieve this acceleration, Microsoft researchers used a new approach, which should also increase the accuracy of speech recognition by 15%.

Using the technology of deep neural networks Microsoft researchers claim that users who rely on speech recognition when writing messages and searches through Bing get results twice as fast compared to the previously used technology. Microsoft declined to compare their technology with solutions from other companies, noting that companies often use different approaches for performance measurement.
result
The new technology is already being implemented in April in servers for speech recognition in the U.S., and the entire process of upgrading the servers should end in the next few weeks. For now there is no information when this technology should be available to customers outside the United States.

Tell is what you think in out Forum

No comments

Leave your comment

In reply to Some User