Last month, Microsoft launched the new AI-based noise suppression feature for Teams meetings. Currently, this functionality is only supported in the Teams Windows desktop client, but the company has now confirmed that it will soon be coming to Macs and mobile devices.
In a detailed blog post, Microsoft gave a quick overview of how the real-time noise suppression feature uses machine learning to filter out unwanted noise during Teams meetings. Microsoft says that it has created its data set to train the ML model to learn the difference between background noise and clean speech.
“Our new noise suppression feature works by analyzing an individual’s audio feed and uses specially trained deep neural networks to filter out noise and only retain speech. 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,” the company explained.
To be clear, Microsoft says that it has not collected customer data for creating this dataset. Interestingly, the engineers extracted specific scenarios from publicly available data and used crowdsourcing techniques for gathering data. They have optimized the deep learning model for human perception to ensure that the feature could run efficiently on almost all Windows devices.
Furthermore, Microsoft has open-sourced the perceptual quality crowdsourcing framework along with the dataset. To advance the research in this field, the company is inviting developers to help the company to improve its ML model. If you're interested, the Team has hosted two competitions so far, and you can learn more on this page.