Screaming kids, barking dogs, and more are all par for the course in video meetings during the coronavirus pandemic.
Fortunately, tech giants are looking to address these realties of working from home by using artificial intelligence (AI) to block out those inevitable background noises.
Microsoft has finally released its AI-powered noise-suppression feature for Teams video meetings. It’s been a long time in the making. Microsoft showed off its tech in March by canceling out the noise of a bag of potato chips being opened. The rustling was completely gone but the speaker’s voice could still be heard. In October, it promised the feature would come to Teams by November.
SEE: Top 100+ tips for telecommuters and managers (free PDF) (TechRepublic)
November has already passed but the feature arrives ahead of Christmas for anyone managing video meetings from their kitchens, lounge, and make-shift study rooms.
“We are excited to announce that users will have the ability to remove unwelcome background noise during their calls and meetings with our new AI-based noise suppression option,” Microsoft said in a blog post.
Before making a video call, Teams users can go into Settings on the the desktop app by clicking on the user icon on the top right hand side of the app. After selecting Devices from the Settings menu, there should be a Noise Suppression option. Unfortunately for Mac users, this feature is only supported on Teams for Windows 10 right now.
The “high” noise suppression feature aims to block out all background sound that isn’t speech. The default “auto” setting lets the app decide what the optimal level of suppression is based on local noise. Instructions for enabling noise suppression on the desktop on mobile are available on Microsoft’s support page for the feature.
There are a few limitations. The PC’s CPU must support the Intel-hatched Advanced Vector Extensions 2 (AVX2). It’s also not available if the meeting or call is being recorded or live captions is turned on. Enabling the feature also consumes more CPU resources.
SEE: WFH and burnout: How to be a better boss to remote workers
The reason it consumes more power is because the feature analyzes the user’s audio feed and employs deep neural networks to isolate speech from noise.
“While traditional noise suppression algorithms can only address simple stationary noise sources such as a consistent fan noise, our AI-based approach learns the difference between speech and unnecessary noise and is able to suppress various non-stationary noises, such as keyboard typing or food wrapper crunching,” Microsoft explains.
In August, Cisco bought BabbleLabs for its AI tech that helps detect human speech and reduce unwanted noises in video meetings and earlier this month rolled out the feature in Webex.
