Online Tone & Noise Generator (white, pink, brown)

A free online tone generator and noise generator: pure tones from 20 Hz to 20 kHz in four waveforms, white, pink and brown noise, frequency sweeps by driver band, a speaker test for channels and polarity, and a synthesized rhythm track. Everything is generated live in your browser with the Web Audio API — nothing to install, no files to download, no royalties attached to any of it.

It is the signal source you need to test a speaker or a subwoofer, set levels with pink noise, find a rattle, set an amplifier’s gain or tune by ear. Start at a low volume and raise it slowly: a sustained pure tone is far harder on a driver — and on your ears — than music at the same setting.

How to use the tone generator: what each tab does

Tone: frequency, waveform and channel

The slider is logarithmic, so the whole audible range stays usable and one tap moves you a musical interval rather than a fixed number of hertz; type an exact figure in the box when you need one. It opens at 440 Hz, the standard reference pitch (A4, ISO 16). The sine is a single frequency and the only honest choice when you are testing “can this speaker do 40 Hz?”; square and sawtooth add a long series of harmonics, which makes them useful for hearing distortion and buzzing but hard on tweeters at high level. The channel selector (left / right / both) is how you find out which speaker is actually connected to which output.

Noise: white, pink or brown

As a pink noise generator this is the setting you will use most. Pink noise is the standard because octaves are how we hear: it puts the same energy in 100–200 Hz as in 10–20 kHz, so a flat-ish system reproduces it as a balanced hiss and an analyzer shows a flat line per band. White noise on the same system looks and sounds top-heavy — nothing is broken, that is just what “equal energy per hertz” does over a logarithmic ear.

Sweep: one band per driver

A sweep glides from one frequency to another so you hear the whole range in a few seconds. The bands map to the drivers you are testing:

• Subwoofer 20–80 Hz · Bass (woofer) 20–250 Hz · Mids 250–2000 Hz · Treble (tweeter) 2–20 kHz · Full range 20 Hz–20 kHz, or type your own limits.
• 1 to 30 seconds, logarithmic (equal time per octave, matching the way you hear) or linear (equal time per hertz, which spends most of the run in the treble).
• Once, loop or back-and-forth — the last one is what you want with your ear next to a cabinet, hunting for the exact frequency that buzzes.

The live readout shows the frequency being played, so when something rattles you can read the number instead of guessing it.

Speaker: channels and polarity

The channel test plays pink noise left, right or both, which settles wiring and dead-driver questions in seconds. The polarity test plays noise on both channels and flips one channel’s polarity: with speakers wired correctly, “in phase” sounds centered and full and “out of phase” sounds diffuse, wide and thin in the bass. If you hear the opposite, one speaker has its + and − swapped at some point in the chain — a classic after moving gear or re-terminating cable.

Beat, and the rest of the controls

The Beat tab synthesizes drums and bass with oscillators and noise at 70–170 BPM in four styles, so you can audition a speaker with something musical — kick and bass in the lows, snare in the mids, hi-hats up top — with no copyright attached, because there is no recording involved. Around the tabs you also get a volume slider (it opens at 30 %), an auto-off timer (5, 15, 30 or 60 minutes), saved presets in your browser, and a binaural beat option for the tone tab that sends each ear a slightly different frequency, 1 to 40 Hz apart. Binaural beats only work on headphones — over speakers the two tones simply mix in the air.

What people use a tone generator for

Testing a speaker, a subwoofer or a soundbar

Pick the band that matches the driver, play the sweep at a moderate level and listen for three things: where it stops (a bookshelf speaker going quiet below 60 Hz is normal, not broken), where it buzzes (a rattle that follows the frequency is mechanical — a loose grille, a port, a cabinet panel, a screw), and where it turns harsh (distortion that appears only above a certain volume is the amplifier or the driver running out of headroom). If a channel sounds wrong rather than just limited, the guided speaker diagnosis walks you through the same checks and interprets the answers for you.

Setting amplifier gain in a car

Car audio manufacturers publish test tones for exactly this, and the convention is well established: 50 Hz for a subwoofer amplifier or any low-pass channel, 1 kHz for a full-range or 4-channel amp. Before you start, set EQ, bass boost, loudness and any DSP flat, because head-unit processing changes the signal you are matching. With the volume slider at 100 % the sine is generated at full scale (0 dBFS), which is what a “0 dB test tone” means; bring the slider down for tones with 5 or 10 dB of headroom. Then raise the gain until you hear (or measure) the onset of distortion and back it off. Keep the bursts short: a continuous sine at full power is a punishing signal, not a soundtrack.

Matching levels and calibrating a room with pink noise

Pink noise plus a meter is how every calibration starts. Play it through one speaker at a time and read the level at the listening position with the sound level meter until every channel matches; then look at the spectrum to see what the room is doing to that flat signal. A dip that moves when you move the microphone is a room problem, not a speaker problem.

Hunting rattles, resonances and room modes

Run a slow back-and-forth sweep between 20 and 120 Hz and walk around the room. Where a single note booms and the next one disappears, you are hearing a room mode — a standing wave set by the distance between two walls, not a fault in the speaker. Where something buzzes only at one frequency, that is a physical object resonating: a window, a light fitting, a cupboard door, a coin. The live frequency readout tells you exactly which note to chase.

Sleep, focus and tinnitus masking

Pink and brown noise mask intrusive sounds by filling in the gaps between them, which is why they are used for sleep, for concentration and to make tinnitus less noticeable. Two practical cautions, both measurable. First, level: a study of 14 infant sleep machines published in Pediatrics (2014) found that at maximum volume all of them exceeded 50 dB(A) at 30 cm — the recommended limit for hospital nurseries — and three exceeded 85 dB(A), which for a full night is an occupational-grade exposure. Keep the source a couple of meters away, use the auto-off timer, and check the actual level at the pillow with the sound meter rather than trusting the volume slider. Second, if you use noise for tinnitus, it should sit below the tinnitus, not drown it.

Tuning and checking your hearing

440 Hz gives you a reference A for tuning by ear, and the tone tab lets you sound any other note you need. For “how high can I still hear?”, a generator is the crude version of the test: the level is not controlled and your headphones are not calibrated, so use the hearing test with a proper threshold ladder if you want a number that means something.

What volume is safe, and what a tone can break

A sustained tone is harder on equipment than music at the same reading, because of crest factor. Music peaks 10–20 dB above its average level, so a driver’s voice coil gets short bursts and time to cool; a continuous sine peaks only 3 dB above its average, so the same “volume” delivers far more sustained heat. That is why tweeters die on test tones and square waves, not on symphonies. Use short bursts, moderate levels, and keep square and sawtooth waves away from tweeters at high volume.

For your ears, the same logic applies with numbers attached: NIOSH’s limit of 85 dB(A) for eight hours halves for every extra 3 dB, and a pure tone is more fatiguing than broadband sound at the same level because all the energy sits in one place. On headphones, the 60/60 rule and the exposure limits behind it are the short version.

What this generator cannot do

• It does not produce ultrasound. Browser audio runs at 44.1 or 48 kHz, so nothing above 22–24 kHz can exist in the signal at all, and the tone tab stops at 20 kHz. Anything sold as an “ultrasonic” effect is out of reach by construction.
• It will not repel mice, rats or insects. The FTC sent warning letters to more than 60 makers and sellers of ultrasonic pest-control devices in 2001, after six enforcement actions between 1985 and 1997, precisely because the efficacy claims were not backed by evidence — rodents get used to the sound and come back. A phone speaker adds nothing to that.
• It does not “burn in” a speaker in any meaningful way. A suspension does loosen slightly over the first hours of use; the measurable change is tiny and playing a full-scale tone for hours risks far more than it gains.
• The level is not a calibrated output. Full scale is 0 dBFS in the digital domain, but what leaves your speaker depends on the OS volume, the amplifier and any EQ or loudness in the chain. For an absolute figure you need a meter, not a slider.
• A phone speaker is not a test speaker. It barely radiates below ~500 Hz, so bass tests through it tell you about the phone, not about the note. Use real speakers or headphones for anything below the midrange.

Playing a known signal is half the measurement; the other half is seeing what arrives. Open the spectrum analyzer on another device and compare what you send with what comes back.

Frequently asked questions