The phrase “Ok Google” has been particularly useful for Android users who either have their hands preoccupied to do the actual typing or who just think that voice automation is there to make their lives easier. Either way, Google Voice search has been an important part of their mobile experience. In the latest update, they have adapted a new acoustic technology in order to make the process easier, and maybe even a bit enjoyable (for the user at least).
Google revealed on their research blog that they are now using Connectionist Temporal Classification (CTC) and sequence discriminative training techniques, which are extensions of the recurrent neutral networks (RNN) which should make voice identification easier in noisy and crowded areas. If you want to be truly technical about this new technology, you can click on the source link for all the nitty-gritty details.
But for a layman’s understanding of what this all means, the software has better and faster phoneme predictions and so in theory, it will be able to pick out your voice and enjoy “Ok Google” voice assistance, even if you’re in a noisy area.
The new acoustic models can also now tell the difference between similar sounding letters though the use of feedback loops. It allows Google Voice to take in all the letters and sounds, all at once as the word is said in one breath. They are now using a sophisticated gating system, but again, all the technical details can be found in the source link.
SOURCE: Google Research
VIA: Engadget
Easy online work for all. Make $ 5000 to $7000 per month onliiine. Visit My Profileeeeee and check hyperlink.
hntgfh
Allow me to show you a genuine way to earn a lot of extra money by finishing basic tasks from+ your house for few short hours a day — See more info by visiting >MY*&___(DISQUS)*%___ID)