Tone and noise generator

Two cautions worth taking seriously before either side of this. First, sustained pure tones damage hearing at volumes that do not feel dangerous, because there is none of the variation that normally makes loudness obvious; start quiet and stay there. Second, what you hear at the extremes is your speakers as much as your ears: most laptop speakers produce essentially nothing below 150 Hz and phone speakers even less, so silence at 40 Hz usually means the hardware, not you. Hearing above 16 kHz declines steadily with age and most adults over forty hear nothing above about 15 kHz, which is ordinary ageing and not a problem.

The waveform is the part people skip and it changes what the test tells you. A sine wave is one frequency and nothing else, which is what you want for tuning or for finding the top of your hearing. A square or sawtooth carries a stack of harmonics above the fundamental, and those harmonics are what make a loose driver buzz or a cheap amplifier distort audibly — so they are the honest choice when the question is about the equipment and not the ear.

The noise colours are precise terms, not marketing. White noise has equal power at every frequency, which puts a great deal of energy in the high range where hearing is most sensitive; hence the hiss. Pink noise drops 3 dB per octave, giving equal power per octave, which is much closer to how natural sounds are distributed and why it sounds like rain. Brown drops 6 dB per octave and is dominated by the low end. For masking speech through a wall, pink or brown works better than white, because speech energy is low and mid-range and covering it with hiss just adds a second thing to hear.

Both are synthesised in the page, with no file played back. Nothing streams, so a long sweep costs no bandwidth, an hour of noise costs none either, and both keep working with the network off once the page has loaded. The tone changes the moment you move the slider, with no download to wait for. The noise is a two-second buffer of random samples, generated fresh each time you press play and then looped, so no two sessions are the same run of noise but a single session repeats on a two-second cycle — nothing you can pick out in white, and audible in brown, where the energy sits low enough to follow.