Best Practices for Noise Suppression on Microsoft Teams Rooms

In hybrid meetings, audio is the make‑or‑break factor of perceived quality. A single fan drone, HVAC rumble, clattering cups, or hallway chatter can derail an otherwise great experience for remote participants. Microsoft Teams uses machine‑learning‑based noise suppression to minimize non‑speech sounds across platforms—including Teams Rooms on Windows (MTRW) and Teams Rooms on Android (MTRA)—so speech remains clear and natural for everyone.
​
But room systems are different from laptops: they use installed microphones, distributed speakers, and (often) external DSPs, where double processing or room acoustics can hurt results. This blogpost provides best practice guidance on how to configure noise suppression and how to design rooms so the AI has the best possible signal to work with.

How Teams noise suppression works (in short)
Microsoft’s model detects and attenuates non‑speech sources (keyboards, fans, traffic, dogs) while preserving the speaker’s voice. It’s available on desktop, mobile, Teams Rooms on Windows, and more, with adjustable levels like Auto, Low, and High depending on platform and version. Microsoft enabled ML‑based suppression by default after measuring substantial reductions in noise complaints and improvements in call quality. Its important to note that noise suppression is not a substitute for room design. It works best when the signal already has good SNR (signal‑to‑noise ratio) because of proper mic choice/placement and controlled reverberation. Microsoft’s room planning guidance emphasizes selecting the right audio devices per room size and treating acoustics first.

MTR on Windows offers centralized noise suppression configuration via XML (SkypeSettings.xml), device‑level settings, and Pro Management controls. In some app versions, turning off suppression in the UI didn’t fully disable processing until resolved in later releases so its important to reference the release notes for changes that affect audio processing interactions. Below is the screenshot of the Pro Portal settings for enabling/disabling noise suppression

An alternative method of configuring noise suppression on MTR Windows is to use the XML file SkypeSettings.xml. MTRW reads SkypeSettings.xml at startup. You can set a default for noise suppression and deploy it at scale using your preferred management method (Intune, Pro Management file push, etc.). The XML is processed at boot and deleted once applied.Typical values use by customers are:0 = Off, 1 = Auto, 2 = Low, 3 = High.

<SkypeSettings>
  <NoiseSuppressionDefault>1</NoiseSuppressionDefault>
</SkypeSettings>
​
It is also important to note that users can also enable/disable noise suppression during a meeting on MTR Windows using the audio settings menu. this only applies for the duration of the call and will revert back to its default settings once the meeting ends. This is shown in the picture below:

Avoiding double processing with external DSPs
Large rooms often use external DSPs (AEC, NR, automixing). If Teams noise suppression stays “on” while the DSP is already doing heavy lifting, you may hear pumping, speech artifacts, or reduced intelligibility. Experienced integrators note the following:

• Certified DSPs are designed to cooperate with Teams. When uncertified DSPs are used, Teams may still apply its own processing for safety—raising the risk of overlap. Prefer Teams‑certified DSPs and peripherals when possible, and coordinate which component does AEC/NR.
• If you must use an uncertified DSP, expect Teams processing to remain active per Teams’ policy logic; mitigate by dialing Teams suppression down (Low/Auto) or off where appropriate—and verify in listening tests.
• Vendors do publish known‑issue guidance for their bars/DSPs with MTRW (e.g., disable Teams suppression for HP Poly Studio X and G series MTR Android devices to prevent degraded audio). Always check the OEM's release notes and test thoroughly before rolling into production

Incorporating OEM AI Noise‑Block Technologies ( e.g. Poly NoiseBlockAI & Acoustic Fence)
Beyond Microsoft’s built‑in machine‑learning noise suppression, many Teams Rooms deployments benefit from OEM‑level AI audio enhancements that run before the signal reaches Teams. Vendors such as Poly (now HP | Poly) ship advanced noise‑blocking technologies—NoiseBlockAI and Acoustic Fence—inside their certified Teams Rooms devices. These technologies intelligently remove non‑speech noise and restrict pickup to a defined acoustic zone, dramatically improving the input signal that Teams receives.
Poly’s official documentation highlights that Poly NoiseBlockAI prevents distracting noises and side conversations from interrupting meetings by intelligently blocking out disruptive room sounds, while Acoustic Fence creates a virtual audio perimeter that blocks voices or noises occurring outside the meeting area. These capabilities ensure that only intentional speech from inside the defined pickup zone reaches remote participants. It also states that the latest version of NoiseBlockAI v2 also removes reverberations or echo in the room, which is especially useful in meeting rooms surrounded by glass walls or large sound reflective surfaces.

Why use OEM's AI Noise suppression over Teams Rooms built-in?
Room systems face more challenging audio environments than personal devices—open doors, hallway chatter, HVAC noise, clattering chairs. Having OEM technologies operate at the hardware/DSP level directly inside the video bar or room kit provides several advantages:

• Device‑side noise elimination happens before Teams receives audio, so the ML model has a much cleaner input to refine rather than repair.
• OEM DSPs can apply spatial awareness, using multiple microphones to identify where the sound originates — something cloud‑based suppression cannot do alone.
• Boundary enforcement (Acoustic Fence) lets you literally “draw” an allowed audio zone so that only in‑room participants are heard.
• Better suppression of transient noises like rustling papers, keyboards, or chair squeaks, which are notoriously hard for cloud suppression models.
• Reduces need for aggressive Teams suppression, lowering the risk of over‑processing and choppy speech.

For these reasons, it is best practice to always enable OEM's device manufacturers noise suppression technology whenever available while disabling the built-in noise suppression in Teams Rooms. Below is an example of enabling HP Poly's NoiseBlockAI on the StudioX72 MTR Android certified video bar:

Summary: The Golden Rule
Use OEM noise‑blocking (e.g. Poly NoiseBlockAI + Acoustic Fence) as the first line of defense, and leave Teams Noise Suppression disabled. This delivers the cleanest, most stable audio for hybrid participants while preventing the choppy audio that comes from stacking multiple aggressive layers.