Google says its voice search system is now more accurate, especially in noisy places


All Transform 2021 sessions are available on demand now. Look now.

If you’ve noticed recently that Google understands what you say better using voice recognition on your smartphone, you’re not crazy. Google’s voice search has indeed become more precise, thanks to advances in artificial intelligence, the tech company announced today.

“Today, we are pleased to announce that we have built even better acoustic neural network models using Connectionist Time Classification (CTC) and Discriminant Sequence Training techniques,” the members wrote. from the Google Speech team Haşim Sak, Andrew Senior, Kanishka Rao, Françoise Beaufays and Johan Schalkwyk. in one blog post today. “These models are a special extension of Recurrent Neural Networks (RNNs) which are more accurate, especially in noisy environments, and they are incredibly fast!

The new models work in the Google app for iOS and Android, as well as dictation on Android, which works in some third-party apps, the team wrote.

Google has reported improvements in voice search not once but twice this year. Obviously, the company has invested in the underlying technology. RNN are an increasingly popular approach to making deep learning a type of artificial intelligence, and Google is widely regarded as having a deep bench in deep learning.

But Apple and Microsoft, among others, have also worked on improving their speech recognition capabilities. Meanwhile, Facebook is also doing more in the area, having acquired a voice recognition company,, a few months ago.

Speech could become more important as a contribution to web research in the years to come. Andrew Ng of Baidu, who is known for his work on the so-called Google brain, predicted last year that within five years “50% of requests will be for speech or pictures.”

“As well as requiring much lower computational resources, the new models are more accurate, noise resistant and faster at responding to voice search queries – so give it a try and good (voice) search! Wrote Sak, Senior, Rao, Beaufays and Schalkwyk.

Read it full blog post for more details on how the team managed to achieve the new performance gains.


VentureBeat’s mission is to be a digital public place for technical decision-makers to learn about transformative technology and conduct transactions. Our site provides essential information on data technologies and strategies to guide you in managing your organizations. We invite you to become a member of our community, to access:

  • up-to-date information on the topics that interest you
  • our newsletters
  • Closed thought leader content and discounted access to our popular events, such as Transform 2021: Learn more
  • networking features, and more

Become a member

Source link

Leave A Reply

Your email address will not be published.