Free tool

RTA analyzer

A real-time analyzer for tuning a system by ear and eye. Play pink noise through the speakers (from the test signal panel below or the noise generator), point your device's microphone at them, and see where the system is heavy or thin. A VU meter shows the overall level. Nothing is recorded or sent anywhere.

Input

View

Response

Spectrum smoothing

Bands

Peak hold

Spectrogram

Input level–dBFS

Spread 100 Hz–10 kHz–dB

Loudest band–

Peak (3 s hold)–dBFS

Noise floor–dBFS

Low vs high–dB

Crest factor–dB

Loudness (3 s)–LUFS

0 VU =

Cursor –

EQ HPF–Q –

EQ low 100–dBQ –

EQ low mid–dBQ –

EQ mid–dBQ –

EQ high mid–dBQ –

EQ high 10k–dBQ –

Press Start and allow microphone access. Then play pink noise through the system.

Noise

Level (RMS)

-60-45-30-15

-30 dBFS

Turn your amp or speakers down before pressing Play noise. It starts at -30 dBFS. For more noise types and ambience, use the noise generator.

How to read it

Bars add up the energy in each 1/3 octave or octave band, so pink noise through a flat system shows level bars. Bands that stick up are where the system or room is heavy; dips are where it is thin.
Spectrum is a continuous line with finer detail, smoothed to 1/6, 1/12 or 1/24 octave. It is scaled the same way as the bars, so pink noise reads flat on both and they line up in Both. Use it to find the exact frequency of a peak or notch before reaching for a parametric EQ.
The EQ rows suggest settings that would flatten the system, worked out from the 1/3 octave bands while pink noise plays; use Average for the steadiest numbers. Flat is the middle level of the bands from 160 Hz to 6.3 kHz. The HPF (high-pass, or low cut) sits where the bottom end has fallen 10 dB below flat, since below it the speakers only waste power (Off means they reach down to 20–25 Hz). Low is a shelf at 100 Hz and High a shelf at 10 kHz. Low mid (200–500 Hz), Mid (630 Hz–1.6 kHz) and High mid (2–5 kHz) are bells, each at the frequency in its range that is furthest from flat. Each bell's Q comes from how wide the problem is (the run of bands at least half as far off), from 0.5 for a broad hump to 4 for a narrow notch; the HPF and shelves read Q 0.7. Anything under 1 dB reads 0. Cuts and boosts are capped at 12 dB. They are a starting point: a phone or laptop mic rolls off the extreme lows and highs on its own, so trust the HPF and the shelves less with one, and prefer cutting to boosting. EQ: Vocals (the EQ button under the EQ row steps through Flat, Music, Vocals, Feedback, the default, and None, which hides the EQ row) leans on cuts rather than boosts, since every dB of boost is a dB closer to feedback: the HPF sits at 80 Hz or higher, the low shelf only cuts, the high shelf lifts at most 2 dB, bells lift at most 3 dB and the low mid cuts no more than 4 dB so the voice keeps its warmth. Peaks are dealt with before dips. Anything ringing on the FB strip takes over its bell as a narrow Q 8 notch (shown as EQ FB and the frequency), 3 to 9 dB deep by how hard it rings. EQ: Music aims for a house curve instead of dead flat, the way most people like recorded music on a PA: the lows about 4 dB up below 100 Hz and the top easing off 2 dB past 8 kHz, with the shelves lifting at most 6 dB (4 dB on top) and bells at most 4 dB. EQ: Feedback only ever cuts: peaks are pulled down, dips are left alone, no shelf or bell is boosted, and anything ringing on the FB strip gets the same narrow notch as in Vocals. It also updates about four times faster than the other modes so it keeps up with a ring as it moves. Flat just flattens the response.
Peak hold marks the highest level each band or point has reached. With 3 s a peak stays for 3 seconds and then drops back to the current level; Hold keeps it until you press Reset. Peak hold, the CLIP light, the averaging and the long-term loudness start 2 seconds after Start, so the jump as the microphone switches on and sets its level isn't caught.
Fast and Slow are the usual meter speeds (about 1/8 and 1 second). Average keeps adding up from the moment you pick it or press Reset, which gives the steadiest picture of a system with pink noise.
Spread is the gap between the highest and lowest 1/3 octave band from 100 Hz to 10 kHz. Under about 6 dB is a well-balanced system.
Low vs high compares the low end (63–250 Hz) with the top end (2–8 kHz) using the 1/3 octave bands. Near 0 dB with pink noise is a flat system; +6 dB or more is bass-heavy or dull, below -6 dB is thin or bright. Many engineers like a gentle +3 to +6 dB for speech and music in a room.
Crest factor is how far the peaks sit above the average level. Pink noise reads about 12 dB and live speech or music often 15–20 dB; a small number means the signal is heavily compressed or clipping. Leave at least that much headroom above your average level.
Loudness (3 s) is short-term loudness in LUFS: the last 3 seconds weighted the way broadcast loudness meters weigh them (ITU-R BS.1770), so it tracks how loud the sound seems better than the plain level does. It is measured at the microphone relative to digital full scale, so use it to compare one moment, song or talker with another rather than as a calibrated delivery reading.
LUFS on the VU meter (the third line at the top left in VU view, the right end of the top row in LED view) is the integrated loudness in LUFS since you pressed Start or Reset, gated like a broadcast loudness meter so silences and pauses don't pull it down. It updates once a second; let a whole song or set run to get the number to compare against your target.
Point at the analyzer (or touch it) and the cursor readout at the top of the graph shows the frequency and level under it.
Noise floor is the background level with nothing playing: the level the room stays above 90% of the time over the last 30 seconds (L90), so brief sounds like a cough or a door don't pull it up. Stop the test noise, let it settle for a few seconds and read it. The gap between it and your pink noise level is how much headroom the room gives your measurement; aim for at least 10 dB. It counts the microphone's own hiss too, so phone mics bottom out around -70 to -80 dBFS. Reset starts it over.
The VU meter moves like a classic analog VU needle, rising and falling over about 300 ms, so it shows average loudness. A thin red needle shows the true peak on the same scale, rising instantly and falling 12 dB a second, and the top corners read Avg (the VU average, in dBFS) and Peak (true peak in dBTP: each channel is oversampled 4x as in ITU-R BS.1770 so peaks that fall between samples are caught; it turns red above -1 dBTP, the usual ceiling for streaming). The CLIP light under the Peak reading lights red the moment the signal reaches full scale (0 dBFS, or 0 dBTP between samples) and stays lit until you click or tap it, or press Reset. The needle scale runs from -40 to +15 VU: dark green from -30 to -7, bright green from -7 to -6, yellow from -6 to 0, red to +10 and glowing red from +10 to +15. LED shows the same readings as two segmented bars from -40 to +15 VU, AVG on top and true PEAK below, green up to -6, amber to 0 and red above, with the highest segment of the last 3 seconds left lit. Pick what level reads 0 VU: -20 dBFS matches many US studios, -18 dBFS is the usual broadcast and console alignment, and -14 dBFS suits louder sources. Peak is the highest sample level, held for 3 seconds.
Phase and sound field share the meter screen and need a stereo input, such as a stereo USB interface or the left and right of a console feed; a single phone or laptop mic is mono, reads +1 and shows MONO INPUT. The square on the left (on top on a phone) is the sound field, a goniometer that plots left against right: mono is a straight vertical line, a source panned hard left or right leans along the L or R diagonal, wide stereo spreads into a round cloud, and a line lying along the horizontal means the two sides are out of phase. Its gain follows the level so quiet sources still fill it. In LED view with Field on, the L, R, MONO and SUB columns beside it show the level of each channel on the same scale and colours as the AVG bar: left, right, MONO (the mono sum of the two) and SUB (that mono sum below 120 Hz, as a sub feed off a crossover would carry). After them, ST is short-term loudness (the last 3 seconds) and LT is long-term (integrated) loudness since Start or Reset, both ITU-R BS.1770 LUFS drawn on the same scale with the reading under each column. Phase correlation runs from -1 to +1 over about 300 ms: the Phase reading on the top line beside the averages in LED view (press Field just after LED to show it and the sound field there; it is off by default, and Scope beside it hides the sound field while keeping the phase reading and channel columns), its own needle beside the VU dial in VU view. Near +1 means left and right carry nearly the same signal (mono-compatible), around 0 to +0.5 is normal wide stereo. The scale is blue from +0.5 to +1, white from 0 to +0.5 and red below 0, and each dot in the sound field takes the same colour for its own position, so out-of-phase energy shows up as red dots along the horizontal; red means a lot of the signal is out of phase between the sides and will thin out or cancel when summed to mono. A swapped pair of wires on one speaker or a flipped polarity switch reads steadily near -1. The thin red needle on the dial is the lowest reading of the last 3 seconds.
The thin feedback bar (FB) under the analyzer lines up with its frequency scale. A frequency glows amber, then red, when it shows up as a narrow peak standing well above its neighbours and keeps doing so for about half a second, which is how feedback and room rings look. It fades over a few seconds once the ring stops, so a brief red mark tells you where the system nearly took off. Point at it to read the frequency, then pull that frequency down on your EQ. Pink noise and speech stay dark; a held note or a test tone lights it too, since a sustained pure tone looks the same as a ring.
The spectrogram scrolls from right to left, about 30 pixels a second, with low frequencies at the bottom. Color shows level, from dark blue at 60 dB below the top of the analyzer scale through green and yellow to red at the top. Point at it (or touch it) to read the frequency and how long ago that moment was; the analyzer's cursor line follows. It is the quickest way to spot a tone that comes and goes, a ring that hangs on after the music stops, or hum that sits under everything.
The oscilloscope at the bottom (under the spectrogram) is a live waveform of the last 20 ms or so of the input (the time span is in the bottom right). Each sweep starts where the signal crosses zero going up, so a steady tone stands still. With a stereo input left is white and right is blue, drawn over each other: matching traces mean mono, and one upside down against the other means a polarity flip. The scale follows the level so quiet sources still fill it, and the figure at the top left is the level of the top line in dBFS. Press Oscilloscope under the EQ suggestions to hide or show it.
The test noise plays through this device's speaker or output. Pink is what you tune with: it has equal energy per octave, so it reads flat on the analyzer. White has equal energy per Hz and rises 3 dB per octave on the bars; use it at a low level for tweeters. Brown falls 6 dB per octave for checking subs. Grey follows the ear's equal-loudness curve, so it sounds about as loud at every pitch; on the analyzer it shows a lift in the lows and the top end and a dip around 3 to 4 kHz. The level knob sets the average (RMS) level; peaks run up to about 13 dB higher, so the knob stops at -15 dBFS to avoid clipping. Drag a knob up or down, scroll on it, or use the arrow keys; click the noise knob or a label to change the type.
Touch or hover over the display to read the frequency and level at that point.
Zoom: scroll the mouse wheel (or pinch on a trackpad or phone) over the analyzer to zoom in on the frequencies around the pointer, and drag sideways to move along. The + and − buttons do the same around the middle. Pick 30 or 15 dB to stretch the level scale for small differences, or zoom the level freely (6 to 120 dB): scroll over the dB numbers on the left (or hold Alt and scroll anywhere), or pinch with your fingers one above the other. Once the level is zoomed, drag up or down to look higher or lower, or hold Shift while scrolling. Fit or a double-click goes back to the full 20 Hz–20 kHz, 60 dB view.
The microphone isn't calibrated, so read the shape rather than the numbers. Phone and laptop mics roll off the deep lows and the very top, so treat the bands below about 60 Hz and above 12 kHz as rough. Measure from a few seats rather than one spot.