Room Acoustics Test
Hear where bass pulses smear or boom. MATT is a listening check, not a measured room map. Axial modes are estimates from the dimensions you type, and a clap gives only a rough reverb time. Nothing is uploaded.
1. MATT test: find ringing bass notes
Use your real speakers (and subwoofer), not headphones, and sit in your usual seat. Start quiet and raise the volume until the pulses are clearly audible at normal listening level. Laptop and phone speakers produce little below about 80 to 100 Hz. What you hear includes those speakers. This is a listening check, not a measured room map.
Press Space or M on a keyboard. Marks are shifted back by about 350 ms for a typical reaction, so the hertz can sit a few steps off the moment you heard the smear.
2. Room mode calculator
Hertz figures below are estimates from these dimensions and 343 m/s. They are not peaks measured in the room.
- Cluster near 68.6 Hz: Length 2nd (68.6 Hz), Height 1st (68.6 Hz). Stacked estimates often line up with one boomy note. That is not a measured peak.
- Cluster near 137 Hz: Length 4th (137 Hz), Height 2nd (137 Hz). Stacked estimates often line up with one boomy note. That is not a measured peak.
- Gap 42.9 to 68.6 Hz: no axial estimate falls in this band, so notes here may sound thin at some seats.
| Hz | Estimate | Index | Note |
|---|---|---|---|
| 34.3 | Length 1st | (1,0,0) | |
| 42.9 | Width 1st | (0,1,0) | |
| 68.6 | Length 2nd | (2,0,0) | cluster |
| 68.6 | Height 1st | (0,0,1) | cluster |
| 85.8 | Width 2nd | (0,2,0) | |
| 103 | Length 3rd | (3,0,0) | |
| 129 | Width 3rd | (0,3,0) | |
| 137 | Length 4th | (4,0,0) | cluster |
| 137 | Height 2nd | (0,0,2) | cluster |
| 172 | Width 4th | (0,4,0) | |
| 206 | Height 3rd | (0,0,3) | |
| 274 | Height 4th | (0,0,4) |
Axial estimates use f = 343 × n / (2 × dimension in metres). Tangential estimates use the two-wall form of the same speed of sound. Clusters and gaps are checked on axial estimates only. These are not peaks measured in the room. Openings, soft walls or odd shapes move a real room off the formula.
3. Clap test: rough reverb time (optional, uses the mic)
Stand where you listen, keep the room quiet, press start, wait for "Clap now", then clap once, sharply. The recording stays on this device and the microphone turns off when the pass ends. Nothing is uploaded. A usable decay can produce a rough T20 or T30 estimate. If the decay, the noise floor, or the fit is not usable, the result is a dash, not a number. The mic is uncalibrated, and the phone or laptop is part of the figure. It is not a lab RT60 and not a calibrated dB SPL reading.
The browser is asked to turn off echo cancellation, noise suppression and auto gain. Some phones ignore that request, which can make the decay look shorter than it is.
How to read your room
Why bass sounds uneven
Low notes have wavelengths as long as your room. When a half wavelength fits exactly between two walls, the sound bounces back and forth and builds a standing wave, called a room mode. At the mode frequency the bass is loud at some seats and nearly gone at others. A 5 m long room has its first length mode at 343 / (2 × 5) = 34.3 Hz, then about 69, 103 and 137 Hz. When modes from two dimensions land close together, as in a square room, they stack into one big boomy note. The calculator uses that same formula on the dimensions you type. It does not measure peaks in your room.
Reading MATT smears
MATT plays short tone pulses with gaps between them. In a well behaved room you hear a clean pulse-pulse-pulse. Where the room rings, energy keeps dying away through the gaps, so the pulses blur into a continuous drone or a wobbly hum. That blur is something you hear through your speakers, not a map the browser measured. Mark each spot and compare it with the estimates: a mark near an axial estimate is a hint about which pair of walls to try, not a measured peak. Run the sweep twice and keep marks that repeat.
Fixes, in order of cost
Move things before you buy anything. Avoid sitting exactly halfway along the room, where the first length mode cancels; about 38% from the front or back wall is a common start. Move speakers a little away from walls and corners. For a subwoofer, do the sub crawl: put the sub at your seat, play bass, crawl around the floor, and place the sub where the bass sounds most even. Next, thick bass traps in corners. Last, use EQ to cut peaks. Never boost dips, since a cancellation just eats the extra power. Recheck with the subwoofer test.
Reverb time targets
RT60 is how long a sound takes to fall by 60 dB. For a home listening room or small studio, roughly 0.3 to 0.5 s is a sensible target. Bare rooms with hard floors are often around 0.7 s or more. This clap test may estimate T20 (a fit from about -5 dB to -25 dB, scaled to 60 dB) or T30 (about -5 dB to -35 dB) when the decay, the noise floor, and the fit are usable. If they are not, it shows a dash. The microphone is uncalibrated and includes the phone, so the figure is a rough estimate, not a lab RT60 and not a calibrated dB SPL reading.
Related tools
Check your sub with the subwoofer test, confirm speaker wiring with the stereo test, find how low your system really goes with the bass test, and check how loud you are listening with the decibel meter.
Frequently asked questions
What does the MATT sweep tell me?
It is a listening check, not a measured map of the room. Short tone bursts play through your speakers, so the speaker is part of what you hear. Where the gaps fill in with a drone, mark that spot. The mark is the oscillator frequency, shifted back by about 350 ms for a typical reaction, not a peak a microphone found.
Are the room-mode hertz measured in my room?
No. They are estimates from the length, width and height you type. Axial estimates use 343 m/s and f = 343 × n / (2 × dimension), after feet are converted to metres. Tangential estimates use that same speed of sound across two dimensions. A dimension of 0 is rejected, so the page does not divide by zero or show an infinite hertz figure. Doors, soft walls and odd shapes move a real room off these numbers.
Is the clap result a lab RT60 or a dB SPL reading?
No. A usable recording can produce a rough T20 or T30 estimate from an uncalibrated phone or laptop microphone, and the phone is part of the figure. It is not a calibrated dB SPL level. If the decay, the noise floor, or the line fit is not usable, the page shows a dash instead of a number.
Is anything uploaded, and how do I stop the sound?
Nothing is uploaded. The sweep is generated in the browser, and the clap stays on the device. Stop silences the tone and suspends audio. Leaving the page stops the microphone, sets the output gain to zero, and closes the audio context. Nothing plays until you press Play MATT sweep or Start clap test, which also resumes audio.
Why is a mark not sitting on an estimated mode?
The formula assumes a bare rectangular box. A mark can also be a speaker-boundary effect, or a mode these estimates leave out. A near miss is a hint about which walls to try moving, then listen again. It is not confirmation that a peak was measured.
Should I use headphones, and is volume 0 silent?
Use speakers, and a subwoofer if you have one, from your usual seat. Headphones skip the room, so they will not show bass smear. Phone and laptop speakers are weak below about 80 to 100 Hz. Volume 0 sets the output gain to an exact zero, so the bursts are silent. The slider starts at 20%, and nothing plays when the page loads.