Online Tone Generator

Type a frequency, press play, and the browser synthesises it. Nothing is installed and nothing is recorded.

The frequency is exact. When this page says 440 Hz it is producing 440 Hz: the number comes from the browser's own oscillator, and nothing in the room, the microphone or the hardware can move it.

The level is unknown. The control on this page sets the amplitude of a digital signal. What that becomes by the time it leaves a speaker depends on your system volume, the amplifier behind it, and the driver itself, and a web page can read none of the three. So there is no decibel figure anywhere on this tool, and that absence is a statement about the hardware rather than modesty about the software.

What an exact frequency is good for

Checking that something works. A speaker, a channel, a driver in a pair of headphones. Set the channel to Left, then to Right, and you have separated a dead speaker from a dead cable in about four seconds — or press Mono check and the page does exactly that for you, a second a side and then both. A steady tone is far better at this than music, which is intermittent, stereo, and mostly mid-range — a blown tweeter is inaudible under a bassline and obvious under a tone at eight kilohertz.

Finding out what a room does. Sweep slowly through the bottom two octaves and walk around while it plays. The level will rise and fall as you move, and in a small room it will do so dramatically: at some frequencies the reflections between two parallel walls reinforce each other and at others they cancel. That is why a subwoofer sounds enormous in one corner and thin three feet away, and sweeping is the fastest way to find where in your own room it happens.

Testing the rest of this site. Point the sound spectrum analyzer at a tone from this page and you have a signal with a right answer: the spike should land where you set it, and the note readout should name it. It is the only way to check that an instrument in a browser is doing what it says without owning a second instrument.

Naming a hum you have already found. If the analyser tells you something is sitting at fifty hertz, play fifty hertz here and compare. Matching a tone to a sound by ear is a much finer discrimination than reading two numbers, and it is how you confirm the thing you found is the thing you are hearing.

Two checks that survive an unknown output path

Everything else on this page is careful to say that the level is unknown. The two checks under the settings look like exceptions and are not. Neither asks you how loud anything is; both ask you to hear a difference between two things that went through the same amplifier and the same pair of speakers, and a difference survives an unknown path intact.

Mono check plays the tone on the left for a second, the right for a second, and both for a second, and then puts the channel back where you had it. The step it is on is marked on screen as it plays, so you can watch and listen at the same time. A side that never arrives is a dead speaker, a dead cable, or a balance control someone moved; if the sound stays in one place throughout, both channels are coming out of the same speaker, which is what a mono cable in a stereo socket does.

Phase test flips the right channel upside down. When two speakers are wired the same way round they push the air in the same direction at the same moment, and everything mixed to the middle of the image — which is most of a voice, a kick drum and a bass line — arrives at you twice and adds up. Reverse one of them and those two arrivals cancel instead. So on a correctly wired pair the middle of the sound thins out or disappears when you press it, and comes back when you press it again. Stand between the two speakers and use a low tone: the two arrivals have to be close to equal for them to cancel, which is truest in the middle and at a frequency whose wavelength is long compared with how far you can lean.

What matters is what happens when it does not. If the tone sounds the same either way, or if it sounds fuller with the phase test on, one of your two speakers is already wired backwards — a swapped pair of terminals behind a speaker, or a cable made wrong — and you have been listening to a system with no middle. It is the one fault here that careful listening does not find on its own, because nothing sounds broken: the bass is simply weaker than it should be, everywhere, all the time.

The phase test compares two speakers against each other, so it needs the channel set to Both and is refused while a single side is selected. The mono check sets the channel itself and puts it back, so it runs from wherever you left it. The inversion is part of the tone rather than a mode of the page: it travels in the copied link and it is in the downloaded file, so what you send someone is what you heard.

Why there is no decibel figure on this page

The online decibel meter on this site publishes a number with a band around it: an uncalibrated estimate, typically within about ±10 dB of a reference instrumentKardous & Shaw. That band exists because the input path — microphone, housing, gain — is unknown.

An output path is unknown in exactly the same way, and there is no equivalent of a microphone reading to even estimate it from. The browser hands a signal to the operating system, which applies a volume setting, which reaches an amplifier, which drives a transducer with its own response. Nothing in that chain reports back. Two laptops given an identical signal will produce sound levels far apart, and neither of them knows it.

So the level control here is labelled as a share of full scale, which is the only thing about it that is true. Start it low — especially on headphones, where the transducer is against your ear and there is no room between you and it to lose anything.

This also settles the question people arrive with most often, which is whether a tone from this page can be used to calibrate a meter. It cannot, and not because of any limitation that better software would fix. Calibration means comparing your device against a reference whose output you know; this page is not a reference, because it does not know its own output either. The method that does work — a source you trust, the same spot, a steady thirty seconds, carry the difference forward — is set out under how to calibrate a phone decibel meter.

The four shapes, and when the shape matters

A sine is one frequency and nothing else. It is the right choice for almost everything on this page: checking a speaker, finding a room mode, matching a hum, feeding the analyser. If you are not sure, use a sine.

A square, a triangle and a sawtooth are all the same fundamental plus a stack of harmonics above it, in different proportions. A square keeps the odd harmonics and sounds hollow and reedy; a sawtooth keeps all of them and sounds bright and buzzy; a triangle keeps the odd ones but drops them away quickly, so it sits between a square and a sine.

That matters in two situations. The first is when you want to hear the top of the range through a speaker that cannot reach it: a sawtooth at two hundred hertz puts energy at four hundred, six hundred and eight hundred as well, so you learn something about the whole range from one setting. The second is when you are looking at the analyser rather than listening — a harmonic series on a spectrum is unmistakable, and it is the pattern electrical hum makes, so it is worth knowing what one looks like on a signal you generated deliberately.

The preview draws the ideal shape. What the browser actually plays is built from a finite series of harmonics so that it does not alias, which means a square wave near the top of the range comes out visibly rounder than the drawing. That is the browser being correct rather than the picture being wrong.

Whether you can hear it is not a hearing test

The most popular thing to do with a tone generator is to run the frequency up until it disappears and take the result personally.

Resist that. What you are testing is a chain, and your ears are the last link in it. Laptop speakers roll off steeply at both ends, and most of them cannot usefully produce the bottom octave at all — set this page to twenty hertz through a phone and you will very likely hear nothing, or hear the cabinet rattling rather than the tone. At the top, cheap headphones, Bluetooth codecs and the sample rate itself all take something out before your ears are involved.

So the frequency you stop hearing at is a property of your equipment and your ears together, and this page has no way to separate them. Hearing changes with age and with exposure and it is worth taking seriously; the way to find out about yours is an audiologist with calibrated equipment in a quiet room, not a browser tab. What this page can honestly tell you is a comparison: whether the left ear and the right ear stop at the same place on the same device, on the same day, at the same setting.

Sending a tone to someone else

Two ways out. Copy link puts the whole setting into a URL — the shape, the frequency, the level and the channel — so whoever opens it hears what you were hearing. Download WAV renders ten seconds of it as a file, which is what you want for anything the browser cannot reach: a car stereo, a PA, a device that plays from a memory stick.

The file is rendered through the same audio path that just played the tone, so it carries the same shape rather than a second synthesiser's idea of one — the waveform, the frequency, the channel and the phase test all reach it, because it is the same graph.

Two smaller things worth knowing while the page is open. Space plays and stops the tone while you are not typing in a field, which is this page only: everywhere else on the site that key belongs to the meter. And the tone stops when you switch tabs. It does not pause and it does not resume by itself — this is the only page here that makes a noise, and a noise coming from a tab you are no longer looking at is one you cannot find in order to turn it off. Every setting is still on screen when you come back, so starting it again is one press.

Where this sits on the site

The tone generator makes a sound. The online decibel meter measures one, the analyser breaks one apart, and the decibel chart says what published levels ordinary sounds sit at. This page is the only one of the four that puts something into the room rather than taking it out, which is what makes it useful for testing the other three. It is also why the level control starts low, and why nothing here will tell you how loud it has got.