Google Duo app starts using artificial noise to improve voice calls

Google is rolling out an audio call improving technology for Google Duo that many of us might not have heard before. This tech called WaveNetEQ aims to remove the audio jitter, which you usually hear while making internet calls, by using artificial noise that mimics human speech. This is done using the company's 'Deepmind' neural machine learning platform.

Usually audio jitter happens when packets of voice data don't reach the other party or reach in a non-sequential manner. As per Google, the 99% Duo calls currently witness packet loss. Out of these around 20% lose 3% audio while 10% lose around 8%, which is also a massive number. On the other hand, the Deepmind neural tech is said to have been trained on data coming from over 100 speakers in 48 different languages.

Also read: Forget Houseparty, Google Duo now supports 12 people in a video call

Google says that its new solution is better than the usual method of dealing with the missing packets, which is also called as PLC or Packet Loss Concealment. As Google explains it, "if new audio is not provided continuously, glitches and gaps will be audible, but repeating the same audio over and over is not an ideal solution, as it produces artifacts and reduces the overall quality of the call. The process of dealing with the missing packets is called PLC." It is said that the reciever's PLC module is responsible for creating audio in order to fill the gaps created with missing packets.

Also read: Facebook Messenger gets a desktop version to deal with video-calling surge

The search giant adds that WaveNetEQ will run on the phone itself and can provide "state-of-the-art audio quality and more natural sounding PLC than other systems currently in use."

Although Google has said that this tech has already rolled out to Google Duo, it is not fir sure if all the users have received it or not. Some of the audio examples can be heard in the company's blog post.