Voice Isolation in Microsoft Teams
Voice Isolation in Microsoft Teams
At Enterprise Connect 2024 Microsoft announced a feature called Voice Isolation. Using AI based noise suppression you’ll now be able to isolate your voice away from background noise so you are better heard and understood in calls and meetings.
In order to use voice isolation, a couple of things will need to happen first.
An Admin will need to enable voice enablement for you as a user. This can be done via PowerShell via a Teams Meeting Policy. More information regarding this can be found here.
After that’s complete you will need to create your “voice profile”. Under your Teams Client settings find the Recognition section and create your voice profile. Whichever language your Teams app has set will determine the voice enrolment language and there are currently 25 languages to choose from. If your language isn’t supported Microsoft have said they are “currently exploring fallback options”. Creating your voice profile will require you to read aloud a short paragraph. Which is then stored and used by the voice isolation model to determine your voice and suppress unwanted background noise.
Some questions have sprung to mind with Voice Isolation. So here are some questions and the answers that I could find:
- “Where is my voice data stored?” – Voice data is stored in the Office 365 trusted compliance store and also it is stored and processed in the same region as their Teams data
- “Is this GDPR compliant?” – Yes, Microsoft will take care of the GDPR compliance, based on the tenant’s regions mapping
- “Can admins get access to my recordings?” – Administrators can export the voice and face profiles of users using the Teams Admin Center. This will be a ZIP file that contains the voice sample and face images of the users.
- “Can I delete my voice data?” – At any time you as a user can remove your voice profile and this will delete it
In summary I think this is a great feature and its nice to see AI being used for other purposes rather than content generation and summarisation. We’ve all been on a call with a user we wished would have better audio quality so hopefully this is a good answer to that.
Let me know if the data storage and processing is a concern for you as an end users or if you’ve been asked about this before.