Tone generator

A tone generator for tuning, testing and listening

Play a sine to tune by ear or check your hearing, a square or sawtooth for a buzzier test signal, a sweep to find what rattles in a speaker, pink noise to level a room. The sound is made live by the browser's Web Audio API⁵. The saved file is the same sound rendered offline and written by @audio/encode. Nothing is uploaded.

Notes and hertz

Notes follow twelve-tone equal temperament: a semitone is a frequency ratio of 2^1/12, a cent 2^1/1200. A4 is 440 Hz by ISO 16⁶ and MIDI note 69, middle C is note 60⁷, so C4 is 440 × 2^−9/12 = 261.63 Hz. Set A4 to 415, 432 or 442 Hz and the names follow; the frequency stays as typed. The field shows hundredths of a hertz, and that is exactly what plays.

Additive synthesis

A periodic sound is a sum of sine waves at whole multiples of its frequency. The Additive tab sets the first sixteen. Odd harmonics at 1/n make a square wave, every harmonic at 1/n a sawtooth, odd harmonics at 1/n² with alternating sign a triangle⁵. The organ preset pulls out all nine drawbars of a Hammond organ. They are named in feet after organ pipes, and a drawbar of f feet sounds 16/f times the 16′ pitch: 16′, 5⅓′, 8′, 4′, 2⅔′, 2′, 1⅗′, 1⅓′ and 1′ are harmonics 1, 3, 2, 4, 6, 8, 10, 12 and 16⁸. The browser plays the sum as one oscillator from a PeriodicWave, scaled to a peak of 1 before the level⁵.

Binaural beats

Two tones a few hertz apart, one in each ear, are heard as one tone pulsing at the difference: 245 Hz on the left and 255 Hz on the right beat at 10 Hz. The pulse is not in the air. It comes from the hearing system combining the two ears' signals, first in the superior olivary complex, and it is heard with carriers below about 1000 Hz, best around a few hundred, for differences up to about 30 Hz¹. Every current browser can play it: separate channels through the Web Audio API work in Chrome, Edge, Firefox and Safari¹⁵. The output decides. A mono setting, one earbud or a speaker mixes both tones into both ears, and the binaural beat becomes an ordinary one².

What binaural beats do beyond being heard is not settled. A meta-analysis of 22 studies found a medium overall effect (g = 0.45) on memory, attention, anxiety and pain, depending on the beat frequency, the exposure time and whether it came before or during the task⁹. A systematic review of 14 studies that recorded EEG asked whether brain rhythms follow the beat: five found they did, eight found they did not, one was mixed¹. This page makes no health claim.

Noise and sweeps

White noise has equal power at every frequency. Pink noise falls 3 dB per octave, so every octave carries the same power; here it is white noise through Paul Kellet's filter, accurate to ±0.05 dB above 9.2 Hz at 44.1 kHz¹¹. Brown noise falls 6 dB per octave; here a leaky integrator, flat below 20 Hz instead of rising further, so less of the speaker's travel goes to rumble you cannot hear. Measured on 22 seconds of each from 125 Hz to 8 kHz: 0.0, −3.0 and −6.0 dB per octave.

A logarithmic sweep gives every octave the same time: f(t) = f₀ (f₁/f₀)^t/T, a curve the browser follows sample by sample⁵. It is the swept sine used to measure the impulse response and distortion of speakers and rooms¹⁴. A linear sweep gives every hertz the same time.

Levels and safe listening

The level is the peak in dBFS, decibels below digital full scale. The default, −18 dBFS, is the EBU alignment level, 18 dB below the maximum coding level¹³. A sine's RMS sits 3 dB below its peak. Square, sawtooth and triangle come from the browser's own oscillator, which may drop harmonics to prevent aliasing⁵; measured in Chromium, Firefox and WebKit, their peak then sits up to 2 dB off the level at high frequencies.

dBFS is not loudness: that depends on your volume and your speakers or headphones. The WHO's safe-listening standard allows adults 80 dB for 40 hours a week¹⁰, and a test tone at full volume in headphones can be far above it. The ear is most sensitive between about 2 and 5 kHz⁴, so a sweep at a steady level swells there. The upper limit of hearing falls with age, from the highest frequencies down¹²; a tone you no longer hear still reaches your ears and your speakers.

The same sounds in code

The audio library makes these in a line. A tone is a function of time, audio.from(t => 0.25 * Math.sin(2 * Math.PI * 440 * t), { duration: 2 }); a sweep is a generator. Edit the numbers below and play.

audio.from(8).chirp({
  f0: 20,
  f1: 20000,
  amp: 0.25,
})

References

1. H. Ingendoh, E. S. Posny & A. Heine, “Binaural beats to entrain the brain? A systematic review of the effects of binaural beat stimulation on brain oscillatory activity,” PLOS ONE 18(5) (2023): e0286023, doi:10.1371/journal.pone.0286023. Perception limits after J. C. R. Licklider et al., JASA 22 (1950) and D. R. Perrott & M. A. Nelson, JASA 46 (1969).
2. Mono audio settings: Apple, Mac: Play stereo audio as mono and iPhone: Settings › Accessibility › Audio & Visual › Mono Audio; Microsoft, Windows: Mono audio.
3. R. Plomp & H. J. M. Steeneken, “Effect of phase on the timbre of complex tones,” JASA 46 (1969): 409–421, doi:10.1121/1.1911705: the largest effect of phase on timbre is smaller than tilting the amplitudes by 2 dB per octave.
4. ISO 226:2023, Acoustics – Normal equal-loudness-level contours; R. Nave, Equal loudness curves, HyperPhysics: sensitivity peaks between about 2000 and 5000 Hz, the ear canal resonance.
5. W3C, Web Audio API: OscillatorNode basic waveforms and their Fourier series, PeriodicWave normalization, exponentialRampToValueAtTime.
6. ISO 16:1975, Acoustics – Standard tuning frequency (Standard musical pitch): A4 = 440 Hz.
7. MIDI Association, The MIDI Tuning Standard: middle C is note 60, A440 note 69; J. Wolfe, Note names, MIDI numbers and frequencies, UNSW.
8. Hammond, SK series manual: Drawbars; Electric Druid, drawbar footages and harmonics.
9. M. Garcia-Argibay, M. A. Santed & J. M. Reales, “Efficacy of binaural auditory beats in cognition, anxiety, and pain perception: a meta-analysis,” Psychological Research 83 (2019): 357–372, doi:10.1007/s00426-018-1066-8.
10. WHO, Deafness and hearing loss: safe listening; WHO–ITU, Safe listening devices and systems (ITU-T H.870): 80 dB for 40 hours a week for adults, 75 dB for children.
11. P. Kellet, Filter to make pink noise from white, musicdsp.org (2000).
12. M. N. Kunchur, “The human auditory system and audio,” Applied Acoustics 211 (2023): 109507, doi:10.1016/j.apacoust.2023.109507: age-related loss starts from the high frequencies.
13. EBU, R 68-2000: Alignment level in digital audio production equipment: 18 dB below the maximum coding level.
14. A. Farina, “Simultaneous measurement of impulse response and distortion with a swept-sine technique,” AES 108th Convention (2000), preprint 5093.
15. MDN, ChannelMergerNode and OscillatorNode: Baseline, widely available.