Compare Ethernet with Wi-Fi
Join one test call with the device connected directly to the router.
If the call stabilizes: fix the Wi-Fi path.
Find the cause of frozen video, robotic audio and delayed screen sharing by separating Wi-Fi, upload pressure, router queues, device load, VPN routing and Microsoft service problems.
Video-call troubleshooting guide
Use this guide to identify whether frozen video, robotic audio or delayed screen sharing comes from Wi-Fi, upload pressure, router queues, the meeting device, a corporate VPN or the Microsoft service path.
Start here
A fast download result does not rule out a Teams problem. Calls depend on the outgoing path, packet timing and device workload. Make one comparison at a time so the result identifies the faulty layer instead of creating more variables.
Join one test call with the device connected directly to the router.
If the call stabilizes: fix the Wi-Fi path.
Stop OneDrive, photo backup, CCTV uploads and large file transfers.
If the call stabilizes: upload pressure or bufferbloat is responsible.
During the problem, open More actions → Settings → Call health.
If network metrics stay healthy: investigate the device, app, VPN or remote participant.
Guide index
Identify
This combines the former symptoms and causes sections so you can move directly from what happens in the call to the most useful test.
| What happens in Teams | What it usually means | Best first check |
|---|---|---|
| Other people say your voice is robotic or your video freezes, but you hear them clearly | Outgoing-path trouble. Upload saturation, upstream bufferbloat, Wi-Fi loss or device encoding pressure is damaging the media you send. | Pause uploads, inspect outgoing Call Health metrics and run the LinkSpeed Bufferbloat Test. |
| Incoming audio or video freezes on Wi-Fi but is stable on Ethernet | Local wireless instability. Weak signal, interference, crowded airtime, mesh backhaul or the device adapter is causing jitter or loss. | Keep Ethernet as the control, then work through the unstable Wi-Fi checks. |
| The meeting worsens as soon as you share the screen | Upload or device pressure. Screen sharing adds outbound data plus CPU/GPU encoding work. | Watch Task Manager or Activity Monitor, pause uploads and compare the Teams app with the browser. |
| Calls lag over wired Ethernet only when the work VPN is connected | Corporate routing or inspection. Media may be hair-pinned through a distant VPN gateway, proxy or security appliance. | Where policy permits, compare VPN on and off, then give the result to workplace IT rather than changing the home router. |
| Teams works over a mobile hotspot but not home broadband | Home router, ISP route or local congestion. The alternate connection changed the broadband path and local equipment together. | Retest over home Ethernet with other traffic paused and compare another call service before escalating. |
| All calls and games become unstable whenever the household connection is busy | Bufferbloat or shared upload contention. Real-time packets are waiting behind bulk transfers. | Run the loaded-latency test and check whether the upload phase causes the largest delay. |
| Only one work laptop has the problem while another device works in the same room | Device, client or VDI issue. CPU, memory, drivers, security software, Teams version, a local VPN client or an unoptimised virtual-desktop session is the stronger suspect. | Compare the desktop app, browser and another device. If Teams runs in a virtual workspace, ask IT whether media optimisation is connected or in fallback mode. |
| Several colleagues report the same problem at the same time | Tenant, office-network or Microsoft service issue. A single home broadband change is unlikely to fix a shared event. | Check official service health and ask the organizer or IT team to review Teams call analytics. |
Reference
Use these values as a practical reference, not as a guarantee. Teams adapts video quality to the available connection, and one short burst of loss can be more disruptive than a reasonable average.
| Teams activity | Microsoft recommended upload / download | Best-performance upload / download | Practical advice |
|---|---|---|---|
| Audio meeting | 58 / 58 Kbps | 76 / 76 Kbps | Audio uses little bandwidth but is highly sensitive to jitter, loss and full upload queues. |
| Video meeting | 2.5 / 4 Mbps | 4 / 4 Mbps | Leave extra headroom for other participants, devices and changing Wi-Fi conditions. |
| Screen sharing in a meeting | 2.5 / 2.5 Mbps | 4 / 4 Mbps | Screen sharing can expose a weak upload path or an overloaded laptop even when camera video is acceptable. |
Microsoft publishes these as per-endpoint guidance and notes that actual use changes with layout, resolution and frame rate. See the official Teams network requirements.
| Metric | Stable target | Needs investigation | Likely to affect calls |
|---|---|---|---|
| Round-trip time | Below 200 ms | 200–300 ms | Repeatedly above 300 ms, especially with talk-over delay |
| Jitter | Below 15 ms | 15–30 ms | Above 30 ms or sharp bursts that coincide with robotic audio |
| Received packet loss | 0% or close to zero | 1–2% | Above 2% or consecutive loss bursts |
| Upload headroom | At least 25–30% free during the call | Less than 25% free | Upload repeatedly reaches full capacity |
Use Teams' own live evidence: during a meeting select More actions → Settings → Call health. Microsoft says the panel updates during the call and provides network, audio, screen-sharing and outgoing-video statistics. Open Microsoft's Call Health instructions.
Validate
Follow the checks in order and record each result. Keep the same room, device and meeting where possible so one change produces a useful answer.
Open Teams Call Health while the problem is happening. Select More actions, Settings and Call health. Record round-trip time, received packet loss, send limits and whether the poor stream is incoming or outgoing.
Record a quiet-network baseline. Run the LinkSpeed Ping Test before the call while no one is downloading, uploading or streaming. Note average and maximum delay plus failed requests.
Compare Ethernet with Wi-Fi. If the same call becomes stable over Ethernet, the home broadband line is not the first suspect. Focus on signal, interference, mesh backhaul, drivers or the device adapter.
Pause every avoidable upload. Stop OneDrive, Dropbox, iCloud, Google Photos, CCTV uploads, file transfers and console updates. If other people immediately hear you more clearly, the outgoing path was saturated.
Run the LinkSpeed Bufferbloat Test. A large latency rise during upload or download load shows that router or line queues can delay Teams media whenever the connection is busy.
Check for packet loss. Run the LinkSpeed Packet Loss Test over Ethernet and Wi-Fi. Loss only on Wi-Fi points locally; repeatable wired loss across several services needs wider investigation.
Compare the desktop app, browser and another device. Close heavy apps and watch CPU, memory, GPU and network use. If another device works on the same connection, the original client or hardware is the stronger suspect.
Compare VPN, mobile data and service status. Where workplace policy allows, test once without VPN and once on a mobile hotspot. Check Microsoft 365 service health before escalating.
Browser tests show real web-request behaviour, while the operating system's native ping provides a second path check. Run the router and public tests together for 60–120 seconds while the call is failing.
Find the router address with ipconfig, then replace YOUR_ROUTER_IP.
ipconfig
ping -t YOUR_ROUTER_IP
ping -t 1.1.1.1
Press Ctrl+C to stop and view the summary.
Find the router in Network Settings or with the system route command, then run:
ping YOUR_ROUTER_IP
ping 1.1.1.1
Press Ctrl+C after the problem has occurred.
Router ping also spikes: investigate Wi-Fi, cabling, adapters or the router.
Router stays stable but public ping spikes: retest over Ethernet and investigate the broadband path.
Both stay stable but Teams degrades: focus on Call Health, the device, VPN, Microsoft route or the remote participant.
Managed work laptop still lags with the VPN disconnected? A secure web gateway, endpoint inspection agent or corporate proxy can remain active outside the VPN. Compare a personal or unmanaged device on the same Ethernet connection. If only the managed device fails, send that comparison to workplace IT and do not disable company security software yourself.
Fix
Do not change the broadband package, router, VPN and Teams settings together. Apply one matched fix and repeat the same call or test.
Confirmed when: Call Health and LinkSpeed results improve on Ethernet but degrade on Wi-Fi.
Do this: keep important calls wired where practical, reposition the router, avoid weak extenders, reduce mesh hops and update the wireless driver.
Retest: repeat the same meeting in the same room, then use the unstable Wi-Fi guide for the remaining radio problem.
Confirmed when: outgoing audio or video improves as soon as cloud sync, CCTV or file transfers stop.
Do this: schedule backups outside meeting time, cap upload-heavy apps and leave at least 25–30% upload headroom.
Retest: restart one upload at a time while watching Teams Call Health to identify the trigger.
Confirmed when: idle delay is stable but rises sharply during the bufferbloat test.
Do this: enable correctly configured SQM or sensible QoS where supported and shape slightly below repeatable wired speeds.
Retest: repeat both loaded directions and follow the bufferbloat and SQM guide.
Confirmed when: another device works in the same meeting, room and network.
Do this: update Teams, the operating system and Wi-Fi/audio/video drivers; close heavy apps; compare the desktop app and supported browser. On a managed work device, note any security, secure-web-gateway or proxy agent and give the device comparison to IT rather than disabling company controls.
VDI or virtual workspace: if Teams runs inside Citrix, Azure Virtual Desktop, Windows 365 or Omnissa Horizon, ask IT to confirm that Teams media optimisation is connected rather than using fallback or server-side rendering. Loud fan noise, sustained CPU use, frozen video or screen-share problems can indicate that media processing is not being offloaded efficiently to the endpoint.
Retest: watch CPU, memory and GPU use during camera video and screen sharing separately. On VDI, repeat the call after IT confirms the optimisation state.
Confirmed when: the wired connection is stable without VPN but Call Health degrades when the corporate tunnel is active.
Do this: give workplace IT the comparison and ask them to review Microsoft 365 Optimize routing, proxy inspection and permitted split tunneling.
Retest: use the IT-approved route. Do not bypass required workplace security controls permanently.
Confirmed when: repeatable wired loss or latency affects Teams and another real-time service with household traffic paused.
Do this: replace the test cable, bypass powerline adapters and extra switches, then collect quiet-time and problem-time evidence.
Retest: use the same endpoint and escalate the timestamps, loss and latency results to the provider.
Do not manually port-forward Teams media on a home router. Teams uses outbound UDP 3478 for STUN, 3479 for audio, 3480 for video and 3481 for screen sharing toward Microsoft media endpoints. If a managed firewall, proxy or inspection service blocks the preferred UDP path, Teams can fall back to permitted TCP connectivity, but delay-sensitive media is generally less resilient on that route. This is an IT or network-team configuration, not a home inbound-port rule. Changing DNS also does not normally fix Wi-Fi jitter, upload saturation or bufferbloat. See Microsoft’s Teams media and call-flow guidance.
Escalate
Use IT when the issue follows a managed laptop, VPN, proxy or office network; another home device works; or Teams Call Health degrades only on the corporate route.
Check service health when several colleagues or tenants report the same symptoms, local tests remain healthy or the problem begins suddenly across locations.
Use the provider when Ethernet shows repeatable loss or external latency across several real-time services with household traffic paused and the router gateway remains stable.
Replace the square-bracketed details. Keep the lines that apply to the team receiving the report.
FAQs
Download capacity does not show packet timing or upload stability. Teams can freeze because of jitter, loss, full upload queues or device load while the headline speed remains high.
A Teams call uses upload and download capacity and can expose Wi-Fi contention, router queueing or a VPN bottleneck. A simultaneous speed test also competes with the call, so compare before-call and during-call results over Ethernet and check Teams Call Health before assuming the line has slowed.
Microsoft currently recommends about 2.5 Mbps upload and 4 Mbps download per meeting-video endpoint, with 4 Mbps each way for best performance. Leave additional headroom for the rest of the household.
Upload is the first path to check because the frozen video is what your device is sending. However, Wi-Fi jitter, packet loss, upstream bufferbloat or device encoding load can create the same symptom even when raw upload speed looks adequate.
Screen sharing adds upload traffic and encoding work. Compare upload headroom, loaded latency, CPU/GPU use and the desktop app against the browser.
Yes. It removes most wireless interference and gives you the cleanest control test before blaming the broadband provider or Microsoft.
Yes. A VPN can add distance, congestion and inspection. Give workplace IT a VPN-on versus VPN-off comparison rather than permanently bypassing required security.
Yes. A secure web gateway, endpoint inspection agent or corporate proxy may remain active after the VPN disconnects. Compare an unmanaged device on the same connection and give the result to workplace IT rather than disabling company controls.
Home users should not create inbound port forwards for Teams. DNS rarely fixes robotic audio or freezing; corporate endpoint and firewall changes should be handled by IT.
During a call select More actions, Settings and Call health. Use the live network and media statistics while the problem is occurring.
Contact IT when the issue follows the work device or VPN while independent network tests are stable. Contact the ISP when wired loss or latency affects several services outside Teams.
Supported virtual desktops can offload Teams media to the physical endpoint. If optimisation is disconnected or in fallback mode, the virtual machine may carry more video and screen-sharing work, increasing CPU use and freezing. Ask IT to verify the optimisation state.
Yes. Teams uses outbound UDP 3478–3481 for media connectivity. If a managed firewall or proxy blocks that path, fallback connectivity can reduce call quality. Workplace IT should review it; do not create home-router port forwards.