Microsoft Teams gained new machine learning and artificial intelligence functionality to substantially improve the audio quality of meetings and calls in suboptimal settings.
The functionality addresses ambient noises, poor acoustics and quality for users without a headset. For them, feedback loops are a common occurrence, wherein the sound of a speaker is recorded by a microphone and infinitely looped back and forth. New machine learning functionality solves the problem, as described by AI chief Solomiya Branets in a Microsoft 365 blog.
AI improvements in Microsoft Teams
In addition to feedback loops, the update removes unwanted echoes. This is a welcome feature for anybody that had their train of thought sidetracked by the audio of their own words. Furthermore, Teams now supports full duplex sound, improving communication across the board. Users can talk and listen simultaneously, which allows for interruptions and makes dialogue feel natural. This solves a typical teleconferencing issue in which audio distorts due to one person speaking over the other.
On the acoustic front, Teams now uses AI to decrease reverberation, resulting in better audio quality for users in rooms with bad acoustics. Even in a huge room, where conversation and other noise might bounce from wall to wall, users sound like they’re speaking into a headset mic. The audio produced sounds similar to that of talks held in an office.