Audio signal generator

440
0

Audio reference

How this app produces sound

Audio synthesizes sound inside your browser. It does not use the microphone, play a recording, or send your settings anywhere. Opening this Help page pauses a playing tone. Close Help and press Play when you are ready to listen again.

1. Frequency generation

Frequency and pitch

Frequency is the number of pressure-wave cycles produced each second. Its unit is hertz, abbreviated Hz. A 440 Hz signal completes 440 cycles per second and is heard near the musical pitch A above middle C. Increasing frequency generally raises perceived pitch; decreasing it lowers perceived pitch.

The center-frequency slider covers 20 through 20,000 Hz. It is logarithmic rather than linear: equal distances represent equal frequency ratios. This gives low and middle frequencies useful control space while still reaching the upper end of the nominal human hearing range. The number above the slider is the exact center frequency used by the synthesizer.

Frequency = 20 × (20,000 ÷ 20)(slider position ÷ 1,000)

The slider has 1,001 internal positions from 0 through 1,000, and the mapped result is rounded to a whole hertz for synthesis, display, and storage. Frequency controls the repetition rate, not loudness. Very low settings may be perceived as pulses or vibration, while high settings may be faint or inaudible. Hearing limits vary substantially between people and commonly change with age.

Wave shape and harmonics

Every active voice uses the selected SINE, SQUARE, or TRIANGLE shape. The center frequency is the fundamental frequency—the lowest repeating rate that defines the pitch. The shape determines which additional harmonics accompany that fundamental.

SINE
A smooth oscillation containing only the fundamental in the ideal case. It has the least complex spectrum and usually sounds the softest or purest.
SQUARE
Alternates between positive and negative levels. It contains strong odd-numbered harmonics at 3, 5, 7, and higher multiples of the fundamental, producing a bright, buzzy sound.
TRIANGLE
Rises and falls in straight lines. It also contains odd-numbered harmonics, but their strength falls much faster than in a square wave, so it sounds smoother and less bright.

Starting, stopping, and routing

Sound begins only after Play is pressed. The browser then creates nine oscillator voices, mixes the audible voices into one mono signal, applies inversion and stereo position, and finally sends the result through a conservative master level to the device output. Play and Pause use short gain ramps so the signal does not begin or end with a sharp digital click. Leaving or hiding the app also stops playback.

INVERT multiplies the entire mixed signal by −1, exchanging every positive pressure direction with a negative one. An isolated signal normally sounds unchanged after inversion. The difference becomes important when it is combined with another related signal because their waves can reinforce or cancel each other.

The bottom slider uses equal-power stereo panning. CENTER sends the mono mix to both channels. LEFT and RIGHT move it to the corresponding channel, while intermediate positions vary the balance smoothly. Panning occurs after all voices have been mixed, so the entire generated sound moves together.

What the top graphic means

The bright center trace identifies the selected wave shape. It always shows four convenient visual cycles; it is not an oscilloscope and does not display the literal number of cycles occurring each second. It remains still while the tone plays. The surrounding traces represent additional voices created by Spread. The dark cutout around the center keeps the original shape readable when many traces overlap.

2. Semitones and voices

What a semitone measures

A semitone is a frequency ratio, not a fixed number of hertz. In twelve-tone equal temperament, one octave is divided into 12 equal ratio steps. Moving up one semitone multiplies frequency by 2 raised to the power 1/12. After 12 upward steps the frequency has doubled; after 12 downward steps it has been halved.

New frequency = starting frequency × 2(semitones ÷ 12)

Because this relationship is logarithmic, a semitone near 100 Hz covers fewer hertz than a semitone near 1,000 Hz, even though both intervals sound like the same musical step.

Spread is the complete width

The SEMITONES value is the total interval from the lowest outer voice to the highest outer voice. The center frequency stays fixed in the middle. Half of the selected interval extends below center and half extends above it.

  • 0 semitones: only the center voice is audible.
  • 12 semitones: a one-octave total span, extending 6 semitones below and 6 above center.
  • 24 semitones: a two-octave total span, extending one octave below and one octave above center.
  • 48 semitones: a four-octave total span, extending two octaves below and two octaves above center.

How the nine voices are placed

The synthesizer has one center voice and four matched pairs around it. Their normalized positions across the total spread are −0.5, −0.375, −0.25, −0.125, 0, +0.125, +0.25, +0.375, and +0.5. Position 0 is the center. Positions −0.5 and +0.5 are the two outer limits.

Voice frequency = center frequency × 2((spread semitones × voice position) ÷ 12)

With a 440 Hz center, 12 semitones of spread reaches approximately 311 to 622 Hz; 24 semitones reaches 220 to 880 Hz; and 48 semitones reaches 110 to 1,760 Hz. The center voice remains exactly 440 Hz in every case.

Level, limits, and what you hear

At zero spread, the eight side voices are silent rather than being stacked on top of the center voice. Their levels fade in during the first 12 semitones. The app then normalizes all audible voice gains so their combined peak ceiling remains comparable to the original single voice. Normalization limits peak level; it does not guarantee that every setting will sound equally loud to human hearing.

Any calculated voice below 20 Hz or above 20,000 Hz is muted, and the remaining voices are renormalized. This means a wide spread near either frequency boundary can become asymmetrical even though the mathematical voice positions remain centered.

Spread creates nine discrete frequency lines, not a continuous band of noise. Nearby voices can produce beating: their phases repeatedly reinforce and oppose one another, causing slow changes in the combined waveform. SQUARE and TRIANGLE settings are even richer because each voice contributes its own harmonic series in addition to the nine fundamentals.

The faint traces in the top graphic use the same voice positions. They separate as SEMITONES increases and disappear when the corresponding audio voice is outside the supported frequency range. The graphic is an explanatory map; the numeric readouts are the exact settings and the generated mix is more complex than any single drawn line.