The Room Mic Capture

Microphone‑measurement tool — Six Walls

The Room Mic Capture is the standalone app used to build the microphone models that ship with The Room. It measures a microphone’s frequency response and writes a .mic file you load into The Room’s MIC selector. It runs on its own — no DAW — and needs only an audio interface, a monitor speaker and a mic stand.

How it works

The app uses substitution. You play a swept sine tone through one loudspeaker and record it twice from the same spot: once with a reference microphone, then again with the microphone you want to model. Dividing the second capture by the first cancels everything the two passes share — the loudspeaker, the room, the sweep itself — and what is left is the difference between the two microphones. Take the reference as your “zero” and that difference is the model.

The result is frequency response only. It does not capture distortion, proximity effect, off‑axis behavior, transient response or self‑noise, and it is only as neutral as the reference mic you measured against. Above roughly 10 kHz the result is uncalibrated unless you supply a calibration file (below) — read the top octave as a tendency, not an exact curve.

Installing

Download from the Applications page on sixwalls.net.

Windows — run the installer, or unzip the portable build and run The Room Mic Capture.exe.
macOS — drag The Room Mic Capture.app to Applications. The build is unsigned: on macOS 15 and later, double‑click it, dismiss the warning, then allow it under System Settings → Privacy & Security (Open Anyway). On macOS 14 and earlier, right‑click it and choose Open.

The app is self‑contained and writes only to your user folder. It shares that folder with The Room, so a model you save here shows up in the plug‑in’s MIC list the next time you open it.

What you need

Reference mic	Your “zero”. A dedicated measurement mic is ideal — with its calibration file it makes the result measurement‑grade. Any mic you trust as a flat baseline works; every model you build is relative to it.
Mic(s) to model	The microphones you want to capture. Switchable mics (pad aside) are worth capturing in each switch position.
One interface	At least one output and one input. One preamp, used for every pass in a session.
One monitor	A single full‑range loudspeaker. Its own response cancels out, so it need not be flat — only steady.
A stand	With a repeatable clamp, so each mic’s capsule lands on exactly the same spot.
A quiet room	Stable temperature, no traffic or HVAC, nobody moving during a sweep.

Setting up

Audio device

The panel at the top is the standard device selector: device type, output device, input device, active channels, sample rate, and buffer size. Pick your interface, enable the input and output channels you will use, and set the sample rate to the interface’s native rate (48 kHz is a good default). If an ASIO entry is offered, prefer it; otherwise Windows Audio (Exclusive Mode) is the next best. Your choice is remembered between launches.

Below the panel, OUTPUT (to monitor) and INPUT (mic) pick the single channels used for the sweep and the recording. Only channels you have enabled in the device panel appear in these lists. If a chosen channel is later disabled, a red warning shows and nothing is played or recorded until you re‑enable it or pick another.

Physical rig — set once, then leave it

Monitor at a fixed distance. Mark the exact capsule position on the stand so every mic returns to the same point.
Working distance about 30–40 cm, on‑axis, identical for every mic.
Orient each mic correctly — end‑address mics pointed at the monitor, side‑address mics with the front of the capsule facing it. Put the capsule, not the body, on the mark.
Set the monitoring level once and do not touch it — not for the reference pass, not for any mic.
Nothing moves between passes: not the stand, not cables, not you.

The capture procedure

Sweep length — 5 s is fine in a quiet room; 8–10 s buys signal‑to‑noise if the room is marginal. Set it once for the whole session.
Set the gain — tick Test tone (1 kHz) and raise the preamp until the meter sits around the −12 dBFS tick. The tone plays at the same level as the sweep, so this is a true preview of the sweep’s recorded level. Untick the tone.
Record reference — mount the reference mic on the mark and click Record reference. One reference pass serves every mic in the session.
(Optional) Load reference cal… — a calibration file for your measurement mic, as a plain‑text list of frequency dB pairs (the format REW and most measurement mics export). The tool divides it out so the model is corrected for the reference mic’s own response. Without one, the result is still valid as a mic‑to‑mic comparison but is uncalibrated in the top octave.
For each mic under test:
1. Swap it onto the mark, in the right orientation.
2. Re‑check the gain with the test tone — a quieter or louder mic is expected and fine; a constant level difference cancels in the maths.
3. Type a name for it.
4. Record mic, then Build profile.
5. Check the curve (next section). If it looks right, Save .mic….
Group the mics by phantom‑power state so you are not switching 48 V on and off through the session — do all the condensers together, then all the dynamics (or vice versa).

Reading the curve

The plot is response in dB against log frequency. 0 dB is the mic’s own average over 400 Hz–2 kHz — the tool normalizes there, so the curve shows shape, not sensitivity. The display clamps at ±18 dB.

What you see	What it means
Gentle rises and dips, a few dB, a soft roll‑off up top	Normal. This is the mic’s voicing.
A ragged patch of ±6–10 dB that repeats when you re‑seat the mic	Real character, common on vintage mics. Keep it.
A large overall tilt, a deep narrow notch, or the ±18 dB clamp hit across a wide band	Something went wrong — the mic moved, the input clipped, the wrong channel was recorded, or phantom power was on/off when it should not have been. Redo that pass.
`build failed`	A capture was silent or far too short. Check the input channel and the meter, then re‑record.

Where the files go

Raw sweep recordings are kept as 24‑bit WAVs in %APPDATA%\The Room\mic-captures\ — reference.wav and one <name>.wav per mic. Keep them: you can rebuild any model from them later (for example after you obtain a calibration file) without re‑capturing.

Save .mic… writes to %APPDATA%\The Room\mics\ by default. A .mic is a mono 32‑bit‑float WAV holding a short minimum‑phase filter, next to a .mic.json sidecar with the name, your notes, the sample rate, the filter length, whether a calibration file was applied, and the measured curve. Keep the two files together.

On macOS both folders live under ~/Library/Application Support/The Room/.

Using the model in The Room

In The Room, click Import mic… next to the MIC selector and choose your .mic file. It is copied into %APPDATA%\The Room\mics, marked in the list with a ▸, and reloads with any project that used it. The mic folder is scanned once when the plug‑in loads, so restart the plug‑in (or your DAW) to pick up files you just added.

Troubleshooting

Symptom	Fix
Meter barely moves; sits near −90 dBFS with the tone playing	That is converter noise on an unused input. Wrong INPUT channel, or the mic’s channel is not enabled in the device panel.
Red “selected channel is not enabled” warning	Enable that channel in the device panel’s active‑channel list, or choose a channel that is enabled.
Every model comes out with the same steep tilt	The reference pass has gone stale — something moved or the level changed since you recorded it. Record the reference again and redo the mics after it.
ASIO is not in the device‑type list	This build was compiled without ASIO. Use Windows Audio (Exclusive Mode).
The app opens on the wrong interface	Pick it once in the panel; the choice, channels, and sample rate are saved for next time.

Credits & licences

JUCE 8 — built with JUCE (juce.com) under the free JUCE Personal license; the “Made with JUCE” splash is part of that license.
The swept‑sine measurement follows the exponential‑sweep deconvolution method (Farina, 2000).