Diagnosis · room → remote
Remote can't hear the room: diagnosing the half-heard table
"We can't really hear you" is three complaints wearing one coat. Faint, chopped and muffled have different causes and different fixes, and treating them as one problem is why so many rooms get "fixed" twice a month and stay broken. Start by making the remote side describe the symptom precisely — then work this order.
First, get a better symptom
Ask the remote participants one question: "Are we too quiet, are words disappearing, or do we sound like we're in a bucket?" It takes five seconds and it halves the search space, because each answer points at a different layer:
| What remote people say | What's likely happening | Where to look first |
|---|---|---|
| "You're faint / far away / only Dana is audible" | Distance and gain — some voices are outside the anchor's useful pickup range | Seating and anchor placement |
| "Words keep cutting out mid-sentence" | Suppression shaving quiet or overlapping speech | Room-system noise suppression level |
| "You sound muffled / underwater / in a bucket" | Wrong microphone selected, an obstructed mic, or genuine reverberation | Input selection on the anchor |
| "It's fine, then it swells and breathes between sentences" | Automatic gain control hunting for distant voices | Seating first, then suppression level |
| "Only when two people talk at once" | Suppression and gain handling overlap badly | Chairing habits, then suppression level |
One prerequisite before any of it: confirm the room has exactly one open audio path. A room with two live devices produces symptoms that mimic every row in that table, because the echo canceller is busy subtracting the wrong things. If you haven't established the anchor yet, stop here, read conference room echo, and come back once the room is clean. Diagnosing a room with a loop in it is diagnosing noise.
The checklist, in order
- Check which microphone is actually selected. This is first because it's the highest-yield check on the list and the most frequently skipped. A room with attached audio very often runs the whole meeting on a laptop's built-in microphone, because plugging in doesn't change what the app chose. On the anchor, open the meeting app's audio settings and confirm the selected input is the room's device. On a Zoom Room with more than one microphone, the controller lets you select the device before setting anything else — worth confirming it's the one you think.
- Look at what's on top of the microphone. Free, instant, and embarrassingly often the answer: a notebook, a laptop bag, a stack of handouts, somebody's elbow, a tent card. Table mics collect obstructions across a working day the way a windowsill collects mugs.
- Move the people, not the settings. Voice energy arriving at a microphone falls away steeply with distance — that's physics, not a vendor claim, and it's why the person beside the anchor sounds present while the far end of the table sounds like weather. Seat the people who talk most inside the anchor's pickup range and the ones who mostly listen further out. It costs nothing and it beats every toggle on this page.
- Reconsider the anchor's position. Mid-table, not at the head; lid at right angles to the conversation rather than facing one end; nobody seated behind it. If one laptop is carrying a whole room, its placement is the single biggest variable you control — the specifics are in one laptop, whole room.
- Now — and only now — look at suppression. If the complaint is "cutting out" rather than "faint," this is probably your answer, and the section below is for you.
- Drop the room's speaker volume a notch. Counter-intuitive but real: a speaker turned up to fill a long room pushes its own microphone harder and gives the echo canceller more to fight, which costs clarity in the outbound direction. If people in the room can't hear the far end at a moderate setting, move them closer rather than turning it up.
- Add cleanup software on the anchor — if the anchor is a laptop. A processing layer between the microphone and the meeting app is the last software lever, not the first. What it can and can't do for a shared room is written up in Krisp for conference rooms.
Bars show the principle, not a measurement — this page states it runs no lab.
When suppression is the thing eating your voices
Noise suppression is trained to keep speech and discard everything else. That works beautifully on a fan, a corridor, a keyboard. It works less well on speech that doesn't look much like speech to it — a quiet aside from four seats away, the tail of a sentence, two people overlapping. Turn suppression up in a room that isn't actually noisy and you are asking a filter to make a harder judgment than it needs to make.
- Zoom Rooms. Zoom's documentation gives the room's Noise Suppression control three options — Off, Auto (the default) and High, which Zoom frames as being for noisy environments. It's reached on the controller at Settings → Microphone → Noise Suppression, and requires Zoom Rooms and controller software 5.13.0 or higher. Crucially, Zoom also documents that a basic noise-suppression layer runs in the background and cannot be disabled — setting the control to Off removes only the more advanced layer. So a room is never truly unprocessed, and your job is choosing a level rather than deciding whether to have one. Our advice: Auto by default, High only for rooms with genuine, persistent noise, and Off purely as a diagnostic. Full walkthrough: Zoom Rooms audio settings.
- Teams Rooms. Microsoft documents that Teams Rooms on Windows uses Microsoft-developed noise suppression by default, combined with microphone manufacturer processing capabilities — there is no room-level aggressiveness dial of the Zoom kind to walk back. On Teams Rooms on Android, audio enhancement is handled by the device manufacturer and the device firmware, and Microsoft states that audio-processing behavior and available controls can vary by certified OEM device, firmware and room-system configuration. In an Android room, a suppression complaint is an OEM documentation question, not a Teams menu question. Details and the DSP wrinkle: Teams Rooms audio settings.
The diagnostic move, where you have the control: change one level, then hold a real meeting. Suppression complaints are notoriously hard to evaluate in an empty room, because an empty room has nothing to suppress. Change it, live with it for a week, ask the remote people directly.
The free fix nobody writes down
The most effective intervention for a half-heard table isn't in any menu: the chair repeats the room. A question from the far end, a muttered objection, the thing that made everyone laugh — the person chairing restates it toward the anchor before responding. It takes three seconds, it costs nothing, and in rooms where the table genuinely is too long for its microphone it is the only thing that works.
It also solves the problem that no processing addresses: remote participants don't just need to hear the words, they need to know a thing was said at all. A room that answers an unheard question looks, from the far end, like a room having a private meeting. We wrote about that asymmetry in the A.O.B. field note on why the room always blames the remote folks.
Where settings stop
Two honest endpoints. First: a table longer than its microphone's useful reach is a physical fact, and no suppression level, gain setting or software layer makes a distant voice present — it makes it processed, which is a different and lesser thing. Second: a reverberant room, all glass and hard surfaces, sends every voice into the mic smeared with its own reflections, and processing can only tidy the edges of that.
Both of those are gear-and-buildings problems with gear-and-buildings answers, and MidTable doesn't do gear or buildings — not because they don't matter, but because purchasing and acoustics are separate disciplines and we'd be pretending. What we can do is tell you cleanly when you've reached the edge, so you stop spending Tuesdays on menus that were never going to help. Everything before that edge — selection, placement, seating, suppression level, one open audio path, a chair who repeats the room — is genuinely ours, and it covers most rooms.
For the wider map of how hybrid rooms fail in both directions, and the order to work them in, start at hybrid meeting audio. To prevent the drift that causes most repeat incidents, run the sixty-second room check before recurring meetings, and decide the anchor once with the Join Map.
Asked by the half-heard
Why can remote people hear some of us but not others?
Distance. Voice energy reaching a microphone drops off steeply, so the seats near the anchor arrive strong and the far end arrives thin — then automatic gain lifts everything to compensate, raising the room's background noise along with the quiet voices. Move the frequent speakers closer; it beats every setting.
Words cut out mid-sentence. Which setting?
Suspect noise suppression before the microphone. In a Zoom Room, check whether the Noise Suppression control has been set to High in a room that isn't genuinely noisy, and return it to Auto. On Teams Rooms on Windows, Microsoft's suppression is on by default with no aggressiveness dial at room level; on Android rooms the behavior belongs to the OEM firmware.
The room sounds muffled. Is that fixable in software?
Often, yes — the usual culprit is the wrong input being selected, so the meeting is running on a built-in mic instead of the room's device. Check the selection on the anchor, then check nothing is physically on top of the microphone. If both are right and it still sounds like a bucket, you're hearing the room's reverberation, which is physical.
Should we just turn the room speaker up?
Usually the opposite. A loud room speaker feeds its own microphone harder and gives echo cancellation more to fight, which can cost outbound clarity. Bring the volume down a notch and move people closer to the far end's voice instead.
We fixed it last month and it's back. Why?
Because fixes that live in one person's memory don't survive a new starter, a docking station or a firmware update reshuffling the selected input. Write the room's settings on a dated card in the room, and run the sixty-second check before recurring meetings.
Verification note: the Zoom Rooms Noise Suppression options, default, controller path, 5.13.0 version requirement and undisableable basic layer, and Microsoft's statements about Teams Rooms noise suppression on Windows and OEM-handled processing on Android, were read on the vendors' official documentation on 20 September 2026. Distance falloff and automatic-gain behavior are standard audio-engineering principles, presented as principles — MidTable runs no measurement lab and publishes no test results.