Calibrate your room
in 20 minutes.

Most readers calibrate their cinema room by running the AVR’s auto-cal once and trusting it. That gets you 70 to 80% of the way to optimal. The remaining 20 to 30% is what separates a good room from a great one.

The cheapest single improvement most readers can make is also the easiest: run auto-cal three times and average the results. Mic position varies even on a tripod. Background noise varies. The auto-cal’s own internal randomisation varies. A single pass produces a result that’s only loosely repeatable. Three passes, averaged, is the difference between a calibration you can trust and one that happens to have caught the room on a good day.

Twenty minutes, a £90 mic, free software, and a willingness to do something three times instead of once. The rest of this guide is the workflow.

Why measurement first

Auto-cal alone is the lazy answer most affiliate guides recommend. To be plain about it: this guide is not anti- auto-cal. You should run auto-cal — every modern AVR has it, it does most of the work, and it’s genuinely good. The point is to verify what auto-cal does, which means measuring independently with something that’s not the AVR.

The kit is accessible, not professional-only. A miniDSP UMIK-1 calibrated USB measurement microphone — around £90, Amazon UK search. REW — Room EQ Wizard, free, available for Windows, macOS, and Linux. A laptop you already own. Add a tripod for the mic (around £20, Amazon UK search) and you have a measurement rig that costs less than a single power-cable upgrade.

What measurement gets you that auto-cal alone doesn’t: an objective view of your room’s frequency response before any correction is applied, the same view after auto-cal runs, and the difference between the two. That difference is the data you need to decide what to leave alone, what to tweak manually, and what auto-cal got wrong. Without it you’re trusting the AVR’s opinion of itself, which is roughly as reliable as a politician’s self-assessment.

If you have to pick one auto-cal system

Audyssey XT32 is the answer for most readers. It’s widely available — every Denon and Marantz AVR in the catalogue runs it, from the entry-level Denon AVR-X1800H up to the Denon AVC-X3800H and the Marantz Cinema 30. The mobile-app version (Audyssey MultEQ Editor) gives you control over target curves and frequency limits the AVR interface hides; budget the £15 — it pays back in one evening.

The alternatives are real and worth knowing about:

• Dirac Live is technically superior — better impulse-response correction, mixed-phase filtering, more transparent results. It ships on the Marantz Cinema 50 and the higher Marantz / Denon flagships, and the full-bandwidth licence is around £350 on top.
• YPAO on Yamaha — including the Yamaha RX-A8A Aventage — is underrated. Genuinely good measurement, slightly conservative correction.
• ARC Genesis on Anthem is best-of-class for accuracy, but Anthem doesn’t list reliably on Amazon UK and isn’t in the catalogue.

The point is not which auto-cal you pick. The point is that you should not pick the AVR for the auto-cal alone — pick it for the amp section, the connectivity, and the tonal voicing, then live with whichever auto-cal it ships.

Where the mic goes

Primary listening seat only.

You’re calibrating for one chair — the one the attentive listener actually sits in. Multi-position mic placement (the auto-cal default, where the system asks you to move the mic to six or eight positions across the seating area) averages the response across positions and inevitably softens the result at the seat that matters most. Each additional measurement position pulls the calibration further from the primary seat in service of seats you’ll only use when the in-laws visit.

Single-seat calibration gives a sharper, more accurate result for the listener who actually pays for the gear. Practical setup: tripod, mic at ear height when seated (around 1.1m for most people), exact lateral and depth position of the listener’s head — not "roughly the middle of the sofa." Sit in the chair, mark where your head is, place the tripod there, then leave the room so your body doesn’t affect the measurement.

When the auto-cal asks for additional mic positions, you have two options: skip them if the AVR allows, or place the mic within a 30cm radius of the primary position so you’re still calibrating for the same seat with minor variance.

What level to calibrate at

Calibrate at THX reference level: 75dB SPLwith the auto-cal’s internal test tone. This is the standard professional reference, and it’s what every cinema mix is mastered against.

The appealing-but-wrong alternative is to calibrate at "what you actually listen at" — the logic being that the calibration is most accurate at your normal level. The problem: your normal level isn’t a fixed thing. You listen quieter on a Tuesday morning when the kids are asleep, louder on Saturday night with the lights down. Calibrate at one and the response drifts at the other. Reference level is a fixed standard. Calibrate at reference, then listen at -10 to -15dB from reference for normal use, or 0dB (full reference) for the rare nights you can.

A sanity-check SPL meter is worth £15 to confirm the AVR’s internal tone is actually 75dB at the seat — Amazon UK search. The internal level on most AVRs is accurate to within ±2dB, but verifying once means you know.

Crossover: don't trust the 80Hz default

Auto-cal will set a crossover for every speaker. The default it picks is almost always 80Hz, which is the THX standard. The 80Hz default is a starting point — the AVR has to set something, and 80Hz is a defensible compromise — but once you have measurement data you should set per-speaker.

Set crossover by your speaker’s natural roll-off:

• Small bookshelves (4–5″ mid-bass): 90–120Hz. Cross higher than the default — small drivers can’t reproduce much below that without distortion.
• Standard bookshelves (6–6.5″): 80–90Hz. The default is in the right zone; tweak by 5–10Hz based on the measured response.
• Floorstanders: 60–80Hz, sometimes lower. Properly braced floorstanders with 8″ mid-bass drivers reach 40Hz cleanly — set 60Hz and let the mains do their job.

Cross too low and the speaker tries to reproduce frequencies it physically can’t, distortion follows. Cross too high and vocal energy moves to the sub’s physical location, breaking the illusion of a single coherent stage. The subwoofer placement guide covers the same point from the sub’s side; the integration only sounds right when both ends agree.

Trust the measurement, but verify the rest

Here’s the principle that resolves the apparent tension between "trust auto-cal" and "do a manual pass":

Trust the measurement-based EQ that flattens your in-room response. Don’t sculpt the target curve.

Verify and tweak the parameter choices auto-cal makes — distances, levels, individual speaker corrections. Auto-cal often gets these slightly wrong.

On the target curve: leave it flat. The Harman and B&K curves are reasonable presets that some listeners prefer, but most rooms respond better to a flat target than to a sculpted one. Sculpting compounds with the auto-cal’s own corrections and creates double-correction artifacts — you end up with a curve that’s flat-then-rolled-off rather than genuinely accurate. Trust the room measurement.

What auto-cal commonly gets wrong:

• Distances: off by 15–30cm regularly. The mic placement, the time- of-flight algorithm, and the AVR’s clock all contribute small errors that add up. Measure with a tape from the mic position to each driver, enter the values manually.
• Levels: channel-to-channel matching often off by 1–2dB. Use a proper SPL meter (or REW with the UMIK) to verify each channel’s level against reference, adjust by hand.
• EQ below 60Hz: most auto-cal systems over-correct in the modal region, applying boost where the room has a deep nullthat can’t actually be filled. Verify in REW; if the pre-EQ measurement shows a -10dB null at 45Hz, the post-EQ measurement should not show that null filled — if it does, the algorithm is wasting amp headroom on a cancellation it can’t solve.

The three-pass rule

Run auto-cal three times. Average the results. Once isn’t reliable.

The reason: a single auto-cal pass is not a deterministic process. Mic position varies — even on a tripod, by 2–3cm if you bump the stand, more if you don’t. Background noise varies — a fridge cycling, a car outside, the neighbour’s extractor fan all push the noise floor around. The auto-cal’s own algorithm has internal randomisation in its sweep timing and filter design. Run it twice with the mic untouched and you’ll get measurably different EQ curves and slightly different distance / level settings.

Three passes, averaged, gives you a result that smooths out the per-pass variance. The how:

1. Run auto-cal pass one. Save the measurement to the AVR’s memory slot 1 (most modern AVRs have multiple slots).
2. Restart the AVR. Run pass two. Save to slot 2.
3. Restart again. Run pass three. Save to slot 3.
4. Compare distances and levels across the three. If two passes agree to within ±5cm and ±0.5dB, that’s the truth — use those. The outlier pass is the one to discard.
5. For EQ, if you’re using REW alongside, average the three measurement files in REW’s All SPL view; the averaged response is what you compare against the post- correction sweep.

Twenty minutes of mic-on-tripod plus three short auto-cal passes plus a coffee while you compare the results. The difference between trusting one pass and trusting an average of three is the difference between a calibration that holds up to scrutiny and one that doesn’t.