The machine learning is something that is essential for Google, because it improves the functioning of your applications and services. Recently, we had the way in which it worked the sensor of the Pixel 4. The brand does now with the improvements in the audio of the calls Google Duo, who have a new model that helps to improve them, as have explained in your blog.
This new model is called WaveNetEQ and it is a system generative based on the technology of DeepMind. Notable for having the ability to complete the packets lost in the waves of voice, which helps the quality of audio in calls on Google Duo dramatically improve.
Google Duo is even better: video calls with up to 12 peopleGoogle Duo is one of the best applications to make group video calls from your mobile. The app now allows video calls with up to 12.
Improving the calls to the Google Duo
When a call is transmitted over the Internet, the packets have certain problems related to the quality. Can be caused by network delays or excessive fluctuations, which cause it to lose a large amount of content. According to Google, up to 8% of the content is lost in this case. This is what has forced the company to take action.
That’s why, have created WaveNetEQ, which is a PLC system (programmable logic controller) that has been trained with a speech database. This generative model allows to synthesize the waves of voice, even in those cases in which we have lost some parts of it. Avoiding such loss referred to before.
Thanks to the use of neural networks, what Google does is that the signal has continuity in real-time. The most direct consequence of this is that the quality is maintained, it reduces the maximum loss of quality. This model is going to “guess” what you intended to say and will complete the wave with those fragments that are missing, those that have been lost.
In this way, such calls in an application as a Duo are going to maintain a good quality, with good audio. Avoiding these interruptions or loss of quality. To train the model has not been simple, but it is in a state that already has the ability to predict what it will say.
Google Duo in the Pixel 4 are starting already to implement these improvements, so that users will be able to enjoy a better audio and a minor loss of quality in their calls. This model is expected to WaveNetEQ go to reach more devices, because it is an improvement in the application, you will be limited to some models only. Unfortunately, for now there are no known dates for this.
The entry Google improves the sound quality of the calls on the Duo: how to have achieved it first appears in The Android Free.