Microsoft Teams Leads in Audio Quality: Echo Cancellation and Noise Suppression Performance
Effective communication is crucial for businesses, and clear audio is a key part of that. Nothing derails a virtual meeting faster than loud background noise or hearing your own voice echo back from the other side. To understand which collaboration platforms deliver the best audio experience, Microsoft commissioned independent tests focusing on two critical areas: echo cancellation (preventing feedback/echo of voices) and noise suppression (filtering out background noise). These studies, conducted in 2024 and 2025 by quality assurance firm TestDevLab, evaluated Microsoft Teams and three other leading communication apps under real-world conditions. The results are enlightening. All the platforms improved call clarity to some degree, but Microsoft Teams emerged as a clear leader across both echo and noise tests. Teams’ advanced audio processing managed to minimize echo and background noise while preserving voice quality better than its competitors. In the following sections, we break down the high-level findings, testing methods, detailed performance results, and what they mean for business users and decision-makers. High-Level Findings Microsoft Teams leads in audio clarity: Across the board, Microsoft Teams delivered the strongest performance in echo cancellation and noise reduction, achieving the highest scores in both objective measurements and subjective listening test. This means meeting participants on Teams experienced the least echo feedback and the cleanest background audio among the platforms tested. Balanced performance is key: Teams was the only platform that excelled at both suppressing unwanted noise and preserving the speaker’s voice quality, resulting in the top overall audio quality ratings in noisy conditions. In contrast, some competitors were very good at noise removal but slightly muffled the speaker’s voice, while others kept voices clear but left a bit more background noise. Teams struck the best balance between reducing noise and maintaining voice clarity, a combination that users find most satisfying. All platforms reduce distractions to some extent: The good news is that all the tested apps effectively reduce echo and noise compared to an untreated call. In fact, none of the platforms significantly degraded voice clarity – all scored high on preserving the original speech. This means whichever platform you use, modern echo cancellers and noise suppressors are doing their job to keep voices intelligible. However, the degree of effectiveness varies, and that’s where Teams had an edge. Rigorous, real-world testing: The studies used industry-standard test protocols (ITU-T recommendations) with both human listeners and automated metrics to evaluate audio performance. Test scenarios included typical use cases like one person talking at a time (to check echo) and both people speaking simultaneously, as well as various background noises (to test noise suppression). By combining objective algorithms with crowdsourced listening panels, the evaluation captured both measurable performance and real user perception. This comprehensive approach lends credibility to the findings. Challenging scenarios differentiate the platforms: Not all situations are equal – for example, echo cancellation is much harder when there are long delays or when multiple people talk over each other. In these stress-test conditions, some platforms’ signal-processing algorithms struggled, whereas Teams’ solution maintained clarity. The reports note that with varying echo delays, conventional echo cancellers perform sub-optimally, which likely explains why competitors fell behind in certain tests. Teams’ strong results in these tough scenarios suggest a more robust or innovative audio technology under the hood. Testing Methodology Both reports followed a similar rigorous methodology to benchmark audio performance, focusing on user experience in realistic scenarios rather than just lab theory. The evaluations were conducted using open-source datasets and test tools, ensuring transparency and reproducibility. Here’s an overview of how the tests were conducted: Standardized protocols: The echo cancellation evaluation adhered to the ITU-T P.831 standard for listening tests, while the noise suppression evaluation followed ITU-T P.835. These are internationally recognized procedures for assessing telephony audio quality. In practice, this means the tests were carefully designed so that results are fair and comparable across platforms. Open-Source Test Material: The evaluations were based on publicly available datasets from leading academic challenges, ensuring transparency and reproducibility. Echo Cancellation: The tests used the blind test set from the ICASSP 2023 Acoustic Echo Cancellation Challenge, comprising 300 real-world recordings (30–45 seconds each). These clips were specifically designed to stress-test AEC systems, featuring difficult conditions such as long or varying loopback delays, device-level DSP (like built-in noise suppression or AEC), microphone distortions, gain fluctuations, and both stationary and non-stationary near-end noise. Noise Suppression: The noise tests drew from the blind test sets used in the DNS (Deep Noise Suppression) Challenges of 2021 and 2022, resulting in 1,874 unique audio clips totaling nearly 7 hours of audio. These samples covered a wide range of acoustic conditions, speaker types, and microphone distances. The test set includes over 150 types of background noise, with variations in speech volume, noise level, language, speaker age, gender, and ethnicity - capturing the diversity of real-world communication environments. Subjective listening tests: For each study, hundreds of audio samples were recorded through four popular communication platforms (Microsoft Teams and three competitors). These samples captured scenarios such as: Near-end single talk: only the local user speaks (no incoming audio) – testing that the speech isn’t falsely removed by echo cancellers. Far-end single talk: only the remote side speaks (the local mic should ideally transmit no echo of that incoming sound back) - a direct test of echo cancellation. Double-talk: both sides speak at the same time - a very challenging scenario for echo cancellers, as they must handle echo and voice simultaneously. Noisy environments: various background noises (e.g., office chatter, traffic, keyboard typing) while people speak - testing noise suppression effectiveness. These recorded clips were then rated by human listeners recruited via Amazon Mechanical Turk (crowdsourced panel). Each listener gave a Mean Opinion Score (MOS) from 1 (bad) to 5 (excellent) on specific aspects of audio quality. In the echo test, listeners rated the amount of echo or other impairments they heard. In the noise test, listeners gave separate scores for speech quality (how clear and natural the voice sounded) and background noise (how well the noise was suppressed) per ITU P.835 guidelines. By averaging these ratings, the researchers obtained comparative scores for each platform under each scenario. Objective measurements: In parallel, the studies employed automated algorithms to score audio quality: For echo, a metric called AECMOS (Acoustic Echo Canceller MOS) was used. This is a machine learning model that predicts MOS scores for echo impairment. It processed the same audio samples to provide an objective check on echo levels and any added distortion. For noise, the DNSMOS model (Deep Noise Suppression MOS) was used. This AI-based metric outputs scores for overall quality (OVRL_MOS), background noise (BAK_MOS), and signal quality (SIG_MOS) of speech. Essentially, it estimates how a typical listener would rate the audio, without needing a human each time. These tools allowed the team to quantify performance in a consistent way. A higher MOS means better quality – e.g., less echo or less noise remaining. Notably, objective scores aligned closely with the human opinions in these tests, boosting confidence in the results. Competitor anonymity: The tested applications included the major players in the virtual meeting market. However, due to legal and competitive concerns, the report anonymized the three non-Microsoft platforms as Competitor A, B, and C. The identities weren’t disclosed, but it’s clear these are well-known communication apps. For our purposes, we’ll refer to them simply as A, B, and C when discussing comparative results. Controls and calibration: To ensure the test was fair, some reference audio samples were included. For example, in the echo study, “ground truth” clips with no echo at all (ideal case) and with very strong echo (worst case) were added to verify that listeners’ ratings made sense. In the noise study, the original unprocessed noisy recordings served as a control baseline (representing how bad it would sound with no noise filtering at all). These controls confirmed the crowd ratings behaved as expected – e.g., the unprocessed audio was rated much worse than any platform-processed call, highlighting how effective modern echo/noise reduction algorithms are. Overall, the testing methodology was comprehensive and real-world. The use of both subjective scores and objective metrics, across multiple scenarios, gives a full picture of performance. For business leaders, this means the findings are grounded in realistic use cases – like your team’s daily Microsoft Teams meeting or a sales call on a conferencing app – rather than abstract lab measurements. Detailed Results Now let’s dig into how each platform performed, first in echo cancellation and then in noise suppression. We’ll see how Microsoft Teams and Competitors A, B, C stacked up in the measured scenarios. Echo Cancellation Results Figure 1: Echo Cancellation performance. Higher MOS indicates better voice quality with less echo. Microsoft Teams achieved the highest score, indicating the least noticeable echo feedback, while Competitor C had the lowest. In the far-end single-talk test – a scenario where only the remote participant is speaking and any echo is purely the platform’s fault – Microsoft Teams had the least echo leakage. As shown above, Teams earned an average listener score of about 3.52 out of 5, the highest of the group. Competitor A was a close second around 3.45. The other two platforms fared worse: for example, Competitor C scored only ~2.9, indicating users heard more echo artifacts on that platform. A difference of even 0.5 on this MOS scale is noticeable to users – so while A wasn’t far behind Teams, the gap between Teams and competitors B and C was significant in this echo test. Perhaps more importantly, Microsoft Teams consistently led across all echo scenarios evaluated. In near-end single-talk (only the local user speaking, which tests if the echo canceller refrains from accidentally cutting out the near-end voice), Teams again received the highest quality rating from listeners, about 3.7 out of 5. Clean reference files were rated around 4, suggesting that Teams’ performance was very close to the best realistically achievable under test conditions. Competing apps’ scores in that scenario ranged from the low 3’s up to 3.5 at best. This implies Teams preserved the speaker’s voice better than others when no echo needed to be canceled (some algorithms can mistakenly dampen the speech even when they shouldn’t – an area where Teams excelled). The double-talk scenario – when both parties speak at the same time – was a major differentiator. Double-talk is notoriously challenging because the system must cancel echo and transmit both voices clearly, all at once. In these tests, Teams outperformed the others notably. It achieved the highest echo-cancellation MOS in double-talk (around 4.6 in objective scoring) with the lowest residual echo, whereas the lowest competitor scored about 4.1. That gap might seem small, but what really set Teams apart was how it handled everything other than echo during double-talk. According to the objective AECMOS analysis, Teams maintained a very high audio quality (scoring 3.8 on the “other” quality MOS), while all three competing apps scored only around 1.7–1.8 on that same scale. In plain language, when people talked over each other, Teams kept the sound clear and stable, but the others introduced many artifacts and dropouts. The report notes that this difference was verified by manual inspection: Teams’ output sounded clean and wasn’t negatively impacted by its echo canceller, whereas the other platforms suffered volume dropouts and glitches during double-talk. This suggests Teams’ echo cancellation algorithm is more robust under pressure, preventing the audio warbling or voice clipping that can occur on other calls when two voices overlap. Another interesting point: the objective measurements (using the AECMOS model) backed up what users reported. In fact, the objective scores indicate Microsoft Teams nearly achieved an “excellent” MOS of ~4.8 in the far-end echo test, whereas competitors ranged between 3.4 and 4.. That aligns with the listener ratings trend we saw. The objective model also flagged that two of the competitors (B and C) had sub-par echo cancellation in the far-end scenario, letting through audible remnants of the far-end speech. Those two also showed the severe double-talk issues mentioned. By contrast, Teams (and to a degree Competitor A) managed to cancel most of the echo and noise from the far end in all scenarios. Overall, the echo cancellation study found that Microsoft Teams delivered the most echo-free experience among the platforms tested. Even in tough conditions – like when someone had a poor speaker/mic setup causing echoes with delay – Teams consistently minimized the echo feedback loop better than others. Its strong performance both in eliminating echo and not distorting the primary speech earned it the top marks in both listener opinions and objective scores. For users, this means calls on Teams are less likely to have that annoying echo that forces everyone to mute and troubleshoot. Competitor A wasn’t far behind in some tests, but Competitors B and C clearly struggled more with echo, especially under challenging conditions. Noise Suppression Results Figure 2: Noise Suppression performance (overall user-rated audio quality in noisy conditions). Higher scores indicate better overall audio clarity after filtering out background noise. Teams achieved the highest overall quality MOS, reflecting an optimal balance of noise removal and voice fidelity, while other platforms trailed slightly. In the background noise tests, all platforms showed huge improvements over the raw, unfiltered audio — but Microsoft Teams again came out on top in overall quality. Test clips for this portion included common noise sources (e.g. café chatter, road noise, typing) mixed with speech. Listeners rated each processed call in terms of how much noise remained (background noise MOS) and how good the voice sounded (speech quality MOS) and an overall quality score, combining the above. Microsoft Teams achieved the highest overall Mean Opinion Score in these noisy scenarios. Microsoft Teams achieved a MOS score of 3.4, clearly placing it in the 'good' to 'very good' range on the 5-point scale. This was the highest among all platforms tested, outperforming Competitor A (3.25), Competitor B (3.2), and Competitor B (3.14). The chart above illustrates this trend: Teams’ bar is the tallest, indicating the best perceived audio in noisy environments, with Competitors A, B, and C following behind. The differences were not gigantic – perhaps on the order of a few tenths of a point – but they were consistent, statistically significant and favored Teams. Why did Teams score highest? The detailed results show that Teams managed to reduce background noise the most without hurting voice quality. In the P.835 listening test, Teams earned the top scores for both MOS_BAK (background noise suppression) and MOS_SIG (signal/voice quality), a rare double win. This led directly to it having the best overall MOS (MOS_OVRL) among the four platforms. In practice, listeners found that a Teams call from a noisy room had less distraction and the speaker’s voice still sounded natural and clear. One or two of the competitors were nearly as good in one dimension but not the other – for example, one competitor matched Teams in noise removal but slightly degraded the voice, and another kept voice quality high but left a bit more noise. Teams was the only one to excel in both dimensions, indicating a more advanced noise suppression algorithm that intelligently distinguishes voice from noise. It’s worth noting that none of the platforms introduced significant voice distortions – all had high MOS_SIG (voice quality scores). The study found that even the original noisy recordings and the processed ones had roughly similar voice clarity ratings. This is by design: modern noise suppression algorithms aim to remove noise without eliminating speech, and they seem to be succeeding. The fact that all platforms scored well on speech quality indicates that business users generally won’t sound robotic or tinny on any of these apps. The main difference comes down to how much noise remains after processing. Here, Teams provided the cleanest audio backdrop. In summary, Microsoft Teams delivered the best audio quality in noisy environments. For a listener, that means a Teams meeting with, say, loud typing or a coffee shop grinder in the background would sound a bit clearer compared to the same situation on other platforms. You would hear less of the unwanted noise, and the person’s voice would come through more naturally. The other top platforms are not far behind – they all make noisy calls far more bearable than no filtering at all – but that extra polish in Teams’ audio can reduce listener fatigue and misunderstandings in a busy workplace. Microsoft’s consistent investment in AI-powered audio stack helps explain these results. Features like real-time noise suppression and AI-based speech enhancement are continuously refined. You can learn more in these blog posts: Reduce background noise in Microsoft Teams meetings with AI-based noise suppression | Microsoft Community Hub New AI-based speech enhancements for Microsoft Teams | Microsoft Community Hub Voice isolation in Microsoft Teams enables personalized noise suppression for calls and meetings | Microsoft Community Hub Conclusion Audio technology in virtual meeting platforms has come a long way, and Microsoft Teams is currently leading the pack. In both echo cancellation and noise suppression, Teams demonstrated superior performance, meaning your conference calls on Teams are likely to have fewer distracting echoes and less background noise compared to calls on other popular platforms. This translates into more productive meetings – participants can focus on the discussion instead of troubleshooting audio issues. That said, the competition is not far behind. The studies showed that all major communication apps are using sophisticated echo and noise control algorithms to improve call quality. In normal conditions, users of any leading platform will enjoy generally clear audio (and certainly far better than an untreated phone call full of echo and noise). The differences emerge in more challenging scenarios – like double-talk and noisy environments. In these edge cases, Microsoft Teams’ advanced processing gave it a noticeable advantage in maintaining clarity. It handled “worst-case” audio situations more gracefully, which can be critical for important meetings. For the industry at large, these results push everyone to continue innovating. After all, as remote and hybrid work persist, expectations for flawless audio will only increase. The good news is that continuous improvements are happening: what was a hard-to-hear echo chamber a few years ago can now sound nearly as good as an in-person conversation. Read the full reports at TestDevLab website: Echo Cancellation Evaluation of Virtual Communication Platforms | TestDevLab Noise Suppression Evaluation of Virtual Communication Platforms | TestDevLab
setReaction and softDelete for channel messages broken?
Hi, I have some issues using the setReaction and softDelete APIs for channel messages in MS Teams. It’s relatively easy to reproduce for me from the the MS Graph explorer. I have a message in a channel, that I programmatically want to set a reaction for (or delete it). The user calling the setReaction API is actually also the owner of the message. This can be seeing by the first screenshot, where updating the message succeeds without problems. But if I either try to set a reaction on the message, or delete it, I’m getting a weird ACL related error message (see second and third screenshot). The fourth screenshot is the relevant part of the decoded access token used by graph explorer, showing that I do have both the ChannelMessage.Send and ChannelMessage.ReadWrite permissions set for the token that are required for the setReaction and softDelete API calls according to the docs. I’ve also tried the same on a chat channel. There setReaction does work as it is supposed to, but softDelete also fails with an error message. Any help here would be appreciated.Compliance recording bot call establishment delay
Hello, I have a compliance recording bot, running for US companies, declared in my tenant. Customers are complaining about long call establishement delays. When looking for Microsoft IP that channel the invites, I notice they are coming from all around the world, from Japan or Ireland for instance. During bot creation I've set up its region to "Gobal", as my tenant allows me only to select "Gobal", "West Europe", and "Central India". I don't know if this is related to this configuration or not. Is there a way to lower these delays, and how ? Thank you for your answer !
What permissions are needed to create teams from a template?
Hi there! I'm trying to create a new classroom team in Teams, but neither my normal account or my GA account has permissions to create teams from templates, nor from another team. The options are greyed out and the tooltip says I don't have permission. The "from scratch" option is completely missing, only "From a group" is available. I've ensured that both accounts have a license (Education A3) and the appropriate permissions in the Teams Admin Portal (where I can create normal teams from scratch with my admin account no problem). Am I missing something? Are there any more permissions required? I've tried both in a browser and in the desktop app.How to Access Participants’ Camera and Mic Device Names in a Teams Plugin
Hello, I’m currently developing an integration for Microsoft Teams focused on enhancing meeting security. One of the features I’m looking to implement involves gathering information about other participants’ camera and microphone device names during or after a meeting. Could anyone point me to the appropriate Microsoft Graph API endpoints, SDKs, or supported methods that allow access to this type of device information for meeting participants? Any guidance on how to implement this functionality correctly and in compliance with Microsoft’s policies would be greatly appreciated. Thank you!
Teams Recap - remove permissions
Hello! Is it possible to remove meeting participants access to a Teams recording and transcription via the Recap? I cannot see any setting to change who has access to the Recap. Is the only way to download then delete the transcription? And change the recording permissions via Stream? TIA!Updates for Town Hall in Microsoft Teams and Teams Live Events
Our goal in Teams is to make hybrid work and communication easier and more inclusive than ever before. This pursuit is core to the effort we put into creating meaningful connections between people through our end-to-end events platforms, whether one-to-one meetings or large one-to-many hosted digital events. We introduced our new digital streaming event solution for large events, town hall, in September 2023. Town hall has continued to drive new, exciting experiences for our customers, such as the ability to bring multiple presenters on stage, send out attendee emails, and see real-time health analytics for the event. As we move forward, we are excited to continue to share our latest features with you and let you know what to expect from town hall in the next year.Introducing Microsoft Frontline Fridays: Your Guide to empowering frontline workers
Are you ready to enhance the way your frontline workforce operates with Microsoft's suite of tools? Join us for Microsoft Frontline Fridays, a series of expert-led sessions designed to help organizations maximize productivity, security, and efficiency for frontline employees. Check out our full series schedule below and register now: Session 1: May 29, 6:00 - 7:30 PM Pacific (UTC-7): Overview & Deployment Best Practices and Strategies (Register Here) Session 1: May 30, 9:00 - 10:30 AM Pacific (UTC-7): Overview & Deployment Best Practices and Strategies (Register Here) Session 2: June 12, 6:00 - 7:30 PM Pacific (UTC-7): Driving successful adoption & delivering value for frontline employees (Register Here) Session 2: June 13, 9:00 - 10:30 AM Pacific (UTC-7): Driving successful adoption & delivering value for frontline employees (Register Here) Session 3: June 26, 6:00 - 7:30 PM Pacific (UTC-7): Streamlining Operational and Corporate Communication (Register Here) Session 3: June 27, 9:00 - 10:30 AM Pacific (UTC-7): Streamlining Operational and Corporate Communication (Register Here) Session 4: July 10, 6:00 - 7:30 PM Pacific (UTC-7): Optimizing Operational Efficiency - How to do more with less, at scale (Register Here) Session 4: July 11, 9:00 - 10:30 AM Pacific (UTC-7): Optimizing Operational Efficiency - How to do more with less, at scale (Register Here) Session 5: July 24, 6:00 - 7:30 PM Pacific (UTC-7): Turbocharging your Employee Home Experience through Customization (Register Here) Session 5: July 25, 9:00 - 10:30 AM Pacific (UTC-7): Turbocharging your Employee Home Experience through Customization (Register Here) Session 6: August 7, 6:00 - 7:30 PM Pacific (UTC-7): Adding agents to your workforce: Transforming frontline efficiency with AI (Register Here) Session 6: August 8, 9:00 - 10:30 AM Pacific (UTC-7): Adding agents to your workforce: Transforming frontline efficiency with AI (Register Here) Why attend Microsoft Frontline Fridays? This series is tailored for organizations looking to explore or optimize their use of Microsoft’s frontline solutions. Key benefits include: Comprehensive Insights: Gain an in-depth understanding of Microsoft’s solutions for frontline workers. Expert Guidance: Learn best practices from Microsoft specialists to maximize the impact of these tools. Interactive Q&A: Get real-time answers from product experts to address your specific challenges. Hands-On Experience: Participate in live demos and interactive activities to build confidence with the platform. What to expect These engaging sessions will provide expert insights, interactive learning, and practical resources to help you deploy and optimize Microsoft’s frontline solutions. Recordings and materials will be shared after each session, and you can choose to attend all or just the ones that interest you. Session Schedule: 1. Microsoft Frontline Solutions: Overview & Deployment Best Practices and Strategies Learn what Microsoft has to offer in the frontline space and how to successfully deploy Microsoft’s frontline solutions with a structured approach. This session covers planning, pilot deployment, environment setup, and app set up to ensure a smooth rollout. Also, learn how to protect frontline workers and organizational data with identity management, secure access controls, device management and secure & compliant solutions that lay the foundation for your tailored frontline environments. 2. Driving successful adoption & delivering value for frontline employees Whether you are shifting from an existing employee engagement solution to Microsoft 365 or expanding M365 experiences to new frontline users or scenarios, managing change in how employees get work done can be the most challenging aspect of technology implementations. In this session, we'll discuss best practices and key considerations related to enabling faster, more successful end user adoption to help your organization get the most from your investments. We'll share real-world learnings and patterns of success based on direct engagement with hundreds of customers across every industry (and we expect we'll learn new things from you too!). 3. Streamlining Operational and Corporate Communication Discover how Teams supports both top-down communications—like community announcements and leadership updates—and local frontline coordination using tools like chat, channels and Walkie Talkie, keeping everyone connected in real time without relying on non-compliant consumer apps. 4. Optimizing Operational Efficiency - How to do more with less, at scale Discover how Microsoft's Frontline solutions can help meaningfully re-engineer your existing business processes to be more efficient, easier to use, and measurable for a digital first and emerging AI world. Learn about everything from off-the-shelf schedule and task management to learning in the flow of work, along with integrating with third party systems of record like your workforce management, inventory and ticketing systems. 5. Turbocharging your Employee Home Experience through Customization Explore rich integrations with third party applications, unique and complex task flows, and home-grown apps, all presented in a single pane of glass. 6. Adding agents to your workforce: Transforming frontline efficiency with AI Learn about how to leverage Microsoft Copilot Chat and agents to optimize frontline workflows - such as relevant, reliable information retrieval, automating repetitive tasks and orchestrating multi-step actions. By participating in Microsoft Frontline Fridays, you’ll gain the tools and insights needed to empower your frontline workforce, enhance efficiency, and secure business operations. Register for the session here.
Teams not linking to document library
When creating a new teams you also create an associated SharePoint site that serves as its document library. This library then gets linked under the files tab in the general channel. At least this is how it is supposed to be. Lately however this does not seem to work correctly anymore The teams and general channel are created correctly, the SharePoint site is also create correctly. The files tab however does no longer link to the correct sub-site on our SharePoint but to the general site who obviously is not a document library, We are temporarily using the work around with creating a new tab and linking to the correct library but this is not a good solution in the long run. I've tested this on different devices, in app and in browser in anonymous browser windows to exclude any client side errors. On the old teams who are working properly clicking on the files tabs send out a get request to the following url <organisation>.sharepoint.com/sites/<teamname>/_layouts/15/filebrowser.aspx?<querystring> On the newly created sited I am not longer getting a reference to which site it should go and we see a url like <organisation>.sharepoint.com/_layouts/15/filebrowser.aspx?<querystring> This doesn't seem to be a fluke as we have tried it on different days, different machines with different accounts. The only thing we have in common is that the Teams admin role is managed trough PIM and has been requested just in time. Any help would be appreciated. The domain or root sharepoint have not been changed since it's creation many years ago
