Speech Enhancement using background noise cancellation.

How many times during an important voice call have you been unable to hear the other person due to traffic noise or loud people in your vicinity? Or how many times have you been in a conference call and had to ask someone to mute themselves or to relocate from a noisy area? Crying babies, barking dogs, and the extremely annoying “hard keyboard typer” all cause you to miss that one piece of important information during the dreaded 2 hour “weekly call”.

We have all been subject to these distractions at one time or another but now, harnessing the power of machine learning, these background noises can be reduced from the call so you can hear conversations more clearly. At SONOC, we have embarked ourselves in a journey to improve speech quality in telecommunications.

Previous models of speech enhancement have issues determining what is a voice and what is noise on the calls, diminishing the success of the voice enhancement. The new more complex method we have created uses deep learning to map the frequency and acoustics of speech vs. noise, and endeavours to filter the “foreign sounds” out of the call.

The inclusion of this feature in our Switching Platform Solution enables the use of this technology in the Wholesale marketplace, helping to improve the quality experience for Carriers. This functionality helps our valued Partners achieve HD-like voice quality on wholesale routes.