You have run your first sweep in Room EQ Wizard, and the line leaps up and down through the bass. Every room does that. Once you know which parts belong to the room and which to the speakers, the graph becomes a to-do list.
What the line shows
The frequency response graph shows loudness in decibels from the deepest bass on the left to the highest treble on the right. It is never the speaker alone: it is the speaker plus every reflection from the room, so the same speaker measures differently at another seat.
Smoothing averages each point with its neighbours, from 1/1 octave, a few gentle hills, to 1/48 octave, every wiggle. Heavy smoothing hides the bass resonances you most need to see; Floyd Toole calls for high resolution in low-frequency room measurements. REW's help recommends its variable smoothing for responses you plan to equalise. To see what smoothing hides in your own file, the REW Analyzer draws a measurement’s 1/12- and 1/3-octave curves side by side, with the unsmoothed one a tick away.
Below about 300 Hz, the room is in charge
In a home-sized room the bass is shaped by room modes, resonances between the room's walls, floor and ceiling. They show as tall peaks and deep, narrow dips that move when the microphone moves. The room mode calculator predicts where they fall for a rectangular room of your size. Toole places the transition, or Schroeder, frequency, below which the room dominates, at roughly 200 to 300 Hz in typical domestic rooms; CEDIA/CTA-RP22, the recommended practice for immersive audio design, gives 150 to 500 Hz for small rooms.
A bass peak is usually a resonance that equalisation can cut. A deep, narrow dip is usually a cancellation: REW's help explains that no amount of EQ restores a true cancellation and that lifting sharp dips often gives poor results, and RP22 lists correcting a room null with amplitude filters among the jobs equalisation is generally not suitable for. Boosting one only wastes headroom; move the seat or the subwoofer instead. For scale, RP22 allows the response below the transition frequency at the main seat to stray from its target curve by at most ±5 dB at its first performance level and ±3 dB at its third, with 1/3-octave smoothing.
Cut the peaks. Move away from the dips.
Shape, decay and ringing
Higher up, judge the overall shape. Toole shows that highly rated loudspeakers in typically reflective rooms produce a curve that slopes gently down from bass to treble, and in double-blind tests by Sean Olive and colleagues, listeners adjusting bass and treble chose a balance close to it, though untrained listeners wanted more bass. He also warns that equalising the wiggles above the transition risks making a good loudspeaker worse.
The RT60 graph shows how long sound takes to fall by 60 dB in each band. RP22 repeats a longstanding guideline of 0.2 to 0.5 seconds for home cinemas: much longer harms speech clarity, much shorter sounds dead. REW's help notes that RT60 is not meaningful below a few hundred hertz in domestic rooms and points to the waterfall for the bass instead: a ridge that runs far back at one frequency is a resonance that keeps ringing. REW’s spectrogram tells the same story as a coloured map of level against frequency and time, which can make a ringing note easier to spot.
Check the measurement, then look closer
A surprising graph is often a measurement problem rather than a room problem. REW subtracts a calibration file only from measurements taken after it is loaded, so check that one was loaded and that it suits the way the microphone pointed: miniDSP’s UMIK-1, for example, comes with a 0° file for pointing at the speaker and a 90° file for pointing at the ceiling, which miniDSP recommends for surround measurements. Compare a before and an after only when they share the same level and settings, and compare a subwoofer’s timing with the speakers’ only when a timing reference, a loopback or an acoustic reference, was used. Our guide to measuring your room with REW covers each of these.
One measurement describes one microphone position. Because room modes and cancellations change from seat to seat, a single sweep cannot prove that a speaker or subwoofer is in the right place, that a panel is needed, or that a crossover or EQ setting is right. Measure the other seats as well, change one thing at a time, and measure again afterwards with the same settings.
When you want help with the reading, open the file in our browser analyzer; the file is not uploaded. It lists the biggest issues first, checks the measurement itself, tells you, once you give it the room's size, whether the lowest bass peaks and dips line up with a single predicted room mode, and overlays a before and an after, or several seats. The response, decay and clarity it shows are recalculated from your file; the room-mode matches, the EQ suggestion and any subwoofer delay are predictions, labelled as such, to test with a new measurement. No measurement yet? Its Sample button loads made-up measurements to practise on. If the bass is the trouble, our subwoofer integration guide takes it from there.



