Your stream sounds like you’re broadcasting from a bathroom, and your viewers have started saying so in chat. Before you spend thousands on acoustic panels, understand what’s actually achievable: a bare, untreated room with a hard floor and drywall will always have *some* reverberation, but you can realistically cut the audible echo your microphone picks up by 50–80% with targeted, mostly free changes. The trick is attacking the problem in the right order — room treatment, then mic placement and processing — instead of throwing plugins at a fundamentally bad source signal.
This guide walks through quick wins first, then a full tuning pass ordered by impact, and finally how to verify your improvements with measurements instead of guessing. Every fix lists what causes the problem, how to check it, what to do, and how to undo it if something sounds worse.
Quick Wins
These changes take under an hour total and produce the fastest audible improvement. Do them before anything else.
| Change | Effort | Expected gain |
|---|---|---|
| Move mic within 4–6 in (10–15 cm) of your mouth, on-axis | Very low | Small to large — the single biggest lever |
| Enable a noise gate plus compressor in OBS (Filters → Audio/Video Filters) | Very low | Small to moderate — masks reverb tails between sentences |
| Drop OBS audio sample rate and boost gate floor correctly (Settings → Audio) | Very low | Small |
| Hang a heavy blanket or moving pad on the wall behind your monitor | Low | Moderate — kills the strongest reflection path |
| Place a rug or carpet remnant on bare flooring under your desk | Low | Small to moderate |
| Switch from an omni-directional mic pattern to cardioid | Low | Moderate if you’re currently omni |
| Reduce NVIDIA Broadcast / RTX Voice “Room Echo Removal” from max to 50–70% | Very low | Moderate with fewer artifacts |
Step-by-Step Tuning
1. Fix mic distance and polar pattern first
What causes it: Every inch between your mouth and the capsule increases the ratio of room reflection to direct signal. A mic 2 feet (60 cm) away captures mostly room; a mic 6 inches (15 cm) away captures mostly you.
How to check: Record 20 seconds of yourself talking, then play it back at low volume. If the reverb tail outlasts your voice noticeably, distance is a major contributor.
What to do: Get the capsule within 4–6 inches (10–15 cm) of your mouth, angled at your lips (on-axis). If your mic is cardioid, point the rear at your noisiest wall or open doorway. Check the physical switch or, on mics like the Shure MV7 or Rode NT-USB, the pattern selector in the companion app. Pair with a boom arm so the distance stays constant — a moving mic makes every downstream setting unstable. Optionally enable the mic’s onboard high-pass filter (usually 75–80 Hz) to cut rumble before compression.
Trade-off: Close miking exaggerates plosives (“p” and “b” pops) and makes mouth clicks audible. Add a foam windscreen or pop filter 2 inches (5 cm) from the capsule. Being this close also means any desk bumps transmit harder — use the boom arm’s isolation, not the desk, as the mount.
How to undo: Slide the boom arm back and flip the pattern switch to its previous position. No software settings change.
2. Treat the wall behind your monitor
What causes it: The surface directly behind your monitor reflects your voice straight back into the front of the mic. Because it’s dead-on-axis for a cardioid mic, this is the strongest single reflection.
How to check: Clap once sharply at your mic position and listen for the fluttering “zing” of parallel hard surfaces. Or walk the wall while talking — you’ll hear the reflection peak directly behind the monitor.
What to do: Hang a moving blanket, heavy duvet, or quilted furniture pad across roughly 4 × 6 ft (1.2 × 1.8 m) of wall behind your setup, mounted on grommets or French cleats so there’s a 2–4 inch (5–10 cm) air gap from the drywall — the gap improves low-mid absorption. If visible blankets bother you on camera, hang them off-frame or cover with decorative fabric panels. Prioritize coverage between 3 and 6 ft (0.9–1.8 m) high, where your mouth and mic sit.
Trade-off: Aesthetics and a small reduction in low-frequency liveliness (your voice may sound slightly “tighter”). Blankets also trap dust, so vacuum them monthly.
How to undo: Take the blanket down and patch the screw holes with spackle.
3. Apply a proper gate → compressor → EQ chain in OBS
What causes it: Reverb tails fill the gaps between your words. A gate silences those gaps; a compressor evens out level so the gate threshold stays predictable.
How to check: In OBS, right-click your mic source → Filters → add “Noise Gate” and watch the level meter while silent. If the meter idles above −50 dB, room noise is high enough that gating helps.
What to do: Add filters in this exact order (drag to reorder in the Filters window):
1. Noise Suppression: set to RNNoise, or NVIDIA Broadcast if you have an RTX GPU.
2. Noise Gate: Close threshold −45 dB, Open threshold −38 dB, Attack 5 ms, Hold 200 ms, Release 120 ms.
3. Compressor: Ratio 3:1, Threshold −18 dB, Attack 6 ms, Release 60 ms, Output gain +3 dB, Knee 6.
4. 3-Band EQ or Expander as needed; cut 200–400 Hz slightly (−2 to −3 dB) if the room sounds boxy.
Trade-off: An aggressive gate clips word endings and makes fast speech sound chopped; a heavy compressor raises the room noise floor between sentences. Tune with headphones while watching the meters, not while distracted.
How to undo: Right-click the filter in the Filters list and remove it, or click the eye icon to bypass without deleting your values.
4. Use AI echo removal — but not at full strength
What causes it: AI-based tools (NVIDIA Broadcast, RTX Voice, AMD Noise Suppression, Krisp) can suppress reverb tails spectrally, but at maximum strength they produce metallic, underwater artifacts.
How to check: Record 30 seconds with the effect on and off, listen on headphones, and A/B for a “phasey” or robotic quality on your voice’s consonants.
What to do: In NVIDIA Broadcast, open your microphone device, set “Room Echo Removal” to 50–70% and toggle “Noise Removal” to about 80%. In Streamlabs Desktop, the “NVIDIA Audio Effects” filter exposes the same controls; in OBS, use the NVIDIA Audio Effects SDK filter if installed. Keep the raw mic as a secondary track (OBS: Settings → Audio → Mic/Auxiliary Audio 2 → select the same device) so you can re-record clean audio later if needed.
Trade-off: GPU cost — expect a few percent of GPU utilization on RTX cards — and artifacts that worsen above 70%. It also can’t recover what was never captured; it suppresses, it doesn’t de-reverberate a bad signal.
How to undo: Set the sliders back to 100%/max or remove the filter.
5. Cover parallel hard surfaces and the floor
What causes it: Flutter echo — a rapid “zipper” ringing — comes from sound bouncing between two parallel bare walls. Hard flooring adds slap-back from below desk height.
How to check: Stand at your mic position and clap. A metallic ring between two walls means those walls are the offenders. Do this for each wall pair.
What to do: Lay a thick rug (at least 5 × 8 ft / 1.5 × 2.4 m, with a dense pad underneath) under and in front of your desk. Break up parallel reflections on the side walls with bookshelves stuffed with books, a fabric couch, curtains (floor-length, pleated — flat panels do little), or first-generation absorption panels with an NRC rating of 0.80 or higher, mounted at ear height at your seated position. A 2-inch (5 cm) panel absorbs speech frequencies well; 4-inch (10 cm) panels reach lower mids.
Trade-off: Cost (quality panels run roughly $30–60 each) and room aesthetics. Over-damping a small room makes it sound unnaturally dead on your end even though the stream sounds better.
How to undo: Remove panels; fill any anchor holes.
6. Cut bass buildup at the corners
What causes it: Corner-loaded bass frequencies (below ~300 Hz) make your voice sound boomy and muddy, which compressors then amplify. This isn’t echo per se, but it makes every echo artifact more obvious.
How to check: Talk while slowly moving into each corner of the room. If the bass swells noticeably, corners are loading up.
What to do: Place thick, loose items in corners: a floor-standing wardrobe full of clothes, laundry hampers of soft goods, or DIY corner traps made from stacked rigid fiberglass panels straddling the corner at 45°. Even a fully loaded closet with the door open helps if the closet sits behind your mic.
Trade-off: Floor space. Bass traps are the least “visible-gain” fix — the improvement shows up as clarity, not less echo, so do it after the steps above.
How to undo: Move the furniture out.
7. Attack desk and monitor reflections
What causes it: Your desk surface and monitor bezels bounce voice energy up into the mic with near-zero delay, creating comb filtering — a hollow, “phasey” tone.
How to check: Record yourself, then listen on headphones for a hollow or flanged quality in your voice, especially when you speak slightly off-axis from the mic.
What to do: Put a desk pad or runner rug on the desk surface between you and the mic. Raise the mic above the desk plane on the boom arm so the capsule isn’t in the bounce path. If you use dual monitors, angle them slightly inward instead of flat-facing each other to break the parallel gap between them.
Trade-off: Minor — a desk pad changes your mouse surface, and raising the mic may put it in the webcam frame.
How to undo: Revert mic height and remove the pad.
8. Re-tune gain staging end to end
What causes it: If your mic gain is cranked to compensate for distance, you’re amplifying the room along with your voice. Gain staging fixes the balance at the source.
How to check: In OBS, right-click the mic meter → “Advanced Audio Properties,” and watch levels while you speak normally. Aim for peaks around −12 to −6 dB, idling around −40 to −50 dB during silence.
What to do: After completing the steps above, lower hardware gain (or the Windows input level in Settings → System → Sound → Input properties) until silence idles near −45 dB, then speak at stream volume and confirm peaks hit −12 to −6 dB. Adjust your Noise Gate thresholds (step 3) to sit just above the new idle level, then re-check every downstream filter.
Trade-off: None really, other than the time to redo the chain — but doing it last matters, since every earlier step changes the noise floor.
How to undo: Restore your previous gain values from memory or a screenshot.
Measure Before and After
Measure at three points: before any changes, after quick wins, and after the full tuning pass. Otherwise you’re relying on memory, which is unreliable for audio.
Method 1 — Clap test recording. At your seated mic position, clap once and record it on your phone held where your mic sits. Count how long the sound takes to die into silence. Shorter decay after treatment = less reverberant energy. Save every recording in a dated folder (e.g., echo-tests\2026-01-15\clap.wav) so you can A/B them later.
Method 2 — OBS filter meter idle level. Open the Noise Gate filter and note the dB level the meter idles at during silence. Lower idle = quieter room. This number should drop several dB between your before and after measurements.
Method 3 — Real playback test. Record two minutes of your actual stream cadence — talking, reacting, keyboard use — through your exact OBS filter chain. Play it back on headphones, then on phone speakers (how many viewers actually listen). Listen specifically for how long the tail rings after each sentence ends. This is the metric your viewers experience.
Repeat the exact same recordings, same mic position, same gain, same time of day (HVAC and street noise vary), for each measurement round. Change only the treatment between rounds.
What Not to Do
Don’t buy thin foam tiles expecting echo reduction. Egg-crate-style foam under about 1 inch (2.5 cm) thick absorbs almost nothing below 500 Hz — the heart of the human voice. It deadens high-frequency slap slightly and does little else. If you buy foam, buy 4-inch (10 cm) thick panels, or skip to blankets and bookshelves for free.
Don’t crank AI echo removal to 100% as a substitute for treatment. Maximum suppression settings introduce metallic artifacts that viewers notice more than moderate echo. It’s a final polish layer, not a fix.
Don’t apply a limiter before compression. The order gate → compressor → limiter matters; a limiter first flattens your dynamics and makes the gate misfire on quiet syllables.
Don’t treat only the wall the mic points at. A cardioid mic’s rear rejection is helpful but not total. Symmetric side reflections still arrive. Spread treatment across surfaces, prioritized by the clap test.
Don’t over-gate. A gate with Close threshold above −35 dB chops trailing consonants (“-s”, “-t”) and makes you sound robotic. If viewers say your words cut off, your Close threshold is too high, not your treatment too weak.
Don’t crank mic gain to “fix” quiet audio. Raising gain raises the room with it. Fix distance and treatment first, then set gain to hit −12 to −6 dB peaks.
FAQ
How much echo reduction is realistic without professional treatment?
In a typical drywall bedroom with hard flooring, expect a dramatic reduction — most of the audible tail gone — from close miking, a treated back wall, and floor coverage alone. You will never reach studio-silence levels, but viewers rarely notice mild room tone when it’s consistent. Full professional treatment mainly buys consistency across the whole room, not just your seated position.
Do acoustic panels behind the mic work better than behind the monitor?
Both matter, but behind and beside your *seated position* (first reflection points) beats directly behind the monitor. Sound travels from your mouth, bounces off nearby surfaces, and re-enters the mic. Map those bounce points: sit in your chair, have someone slide a mirror along the walls — anywhere you can see the mic in the mirror is a first reflection point worth covering.
Should I use noise suppression, a noise gate, or both?
Both, in the right order. RNNoise or AI suppression continuously removes steady background noise (fans, HVAC, hum). A gate mutes everything, including reverb tails, when you’re not speaking. Suppression first cleans what the gate has to judge; the gate then closes more cleanly and quietly. Running only a gate means the room tone “pops in” loudly every time you speak.
Will a dynamic microphone reduce echo more than a condenser?
Usually yes, meaningfully. Dynamics like the Shure SM7B, MV7, or PodMic are less sensitive overall and capture less of the distant room relative to your close voice. A condenser in the same untreated room will pick up more ambience. If you’re buying gear specifically to fight a live room, a dynamic mic is a better investment than foam panels — but it still can’t replace proper distance and treatment.