Test Tone Generator

A sustained note provides a useful reference when matching a pitch, teaching sound synthesis, or repeating an audio exercise under the same conditions. Music changes many things at once. A test tone holds selected properties steady so that a change in frequency, waveform or level is easier to recognize. Amplitude, channel assignment, duration and the way the sound starts and ends matter alongside its frequency.

Frequency counts complete cycles per second and is measured in hertz, or Hz. A 440 Hz tone repeats 440 times each second. Doubling the frequency raises the pitch by one octave, so 440 Hz and 880 Hz are an octave apart. Equal musical intervals follow frequency ratios rather than equal additions of hertz. The step from 100 to 200 Hz spans the same octave interval as the step from 800 to 1,600 Hz.

Waveform describes the shape of each cycle. An ideal sustained sine wave contains one frequency. An ideal symmetric square wave also contains higher frequencies called harmonics, at odd multiples of its fundamental frequency. Those extra components give it a different tone quality even when its repetition rate matches the sine. Browser-generated square waves are bandlimited: the highest harmonics are removed or suppressed, so a generated square does not retain the perfectly sharp edges of its mathematical drawing.

For a sweep, frequency moves continuously between two endpoints. A linear sweep covers equal numbers of hertz in equal time intervals. A logarithmic sweep covers equal frequency ratios, giving each octave the same share of time. This distinction matters when demonstrating pitch relationships or preparing a repeatable frequency sweep for another audio application. Neither curve keeps perceived loudness constant as pitch changes.

Amplitude describes the signal's size. Digital level is often expressed in decibels relative to full scale, written dBFS, with negative values below the digital full-scale reference. Sound pressure level, or SPL, describes acoustic pressure at a measurement point. Device volume, amplification, speakers or headphones, and the listening position all affect the sound that reaches a listener. A digital level therefore cannot establish an acoustic level or a safe listening volume.

Starting or cutting a waveform abruptly can add a click. Short fades bring its amplitude up from silence and back down, reducing that discontinuity. Duration defines the complete sound, including those fades. Channel assignment determines whether a signal occupies the left channel, the right channel or both; identical signals in both channels do not create independent stereo content.

Brief tones can support pitch matching and signal demonstrations, but listening alone does not calibrate equipment or assess hearing. An inaudible tone may still be reaching the audio system. Low device volume and short playback are sensible starting precautions; increasing volume until an extreme frequency becomes audible is not a reliable test.

How to Use This Tool:

Choose a steady reference tone or a sweep before playing. The initial settings are a 440 Hz sine tone on both channels at −30 dBFS for five seconds, with 20 ms fades.

1. Lower your device volume before playback. The Digital level setting does not change the device's volume control. Keep listening brief and stop if the sound is uncomfortable.
2. Choose Sine tone, Square tone or Sine sweep under Signal, then enter the frequency. For a sweep, enter an End frequency and choose a Sweep curve. A lower endpoint makes a descending sweep, while equal endpoints hold a steady sine.
3. Select the Output channel, then set Digital level and Duration. Duration uses whole seconds. Enter fractional frequencies and levels in the number fields; the sliders move in whole units.
   For a different fade duration at both ends, open Advanced and set Fade override to 5 to 100 ms. Blank uses 20 ms, and fractional values are accepted.
4. Check the Configured signal summary before pressing Play tone.
   If Check signal settings appears, correct the field named in its message. Playback stays disabled while an active value is invalid. End frequency and Sweep curve apply only to Sine sweep; returning to a sweep restores those settings and checks them again.
5. Press Play tone to start. Playback ends after the chosen duration, or sooner with Stop / Esc. Editing a setting stops the current sound and requires another Play. Playback also stops when the page becomes hidden or is left; restoring settings from a link never starts sound automatically.
6. Open Tone file and choose Download WAV when you need the configured sound in another audio application. You can download it without playing first. Check the stated signal, channel, duration and fade before reusing the file.

Interpreting Results:

The results describe the planned digital signal. They do not record the speakers, measure room sound, or grade the equipment.

• Signal checkpoints gives five equally spaced times, including the start and end. Left peak (1) and Right peak (1) show the ideal amplitude envelope: the size available to the waveform at that time. The unit “1” means dimensionless. Zero at the endpoints represents the fades, and zero throughout one channel represents an unselected channel. These are not instantaneous waveform samples.
• Frequency path joins those five frequency checkpoints with straight segments. It makes the sweep's direction and progression visible, but the drawn segments are a sparse illustration. Logarithmic playback follows a continuous frequency change between them.
• Waveform preview shows 17 points over one ideal cycle, with phase measured in cycles rather than seconds. It excludes fades and browser bandlimiting. A sweep uses the same ideal sine shape, so this view does not show its changing frequency over the full duration.

Tone file reports an ideal RMS value before fades. Root mean square, or RMS, summarizes signal magnitude over time. Here it is a mathematical waveform reference, not a measurement of the finished file or the sound reaching your ears.

The WAV is always stereo, 48 kHz, 16-bit pulse-code modulation (PCM), with the selected duration and fades. A single-channel selection leaves the other channel silent. Live playback uses the audio sample rate reported in its status, which may differ from 48 kHz. The WAV is newly synthesized audio, not a recording of live playback.

Technical Details:

A digital tone combines a repeating waveform, a frequency trajectory and an amplitude envelope. Frequency determines how quickly the waveform progresses through its cycles. The envelope scales its amplitude through the start, sustained portion and end.

Formula Core

The selected digital level becomes a linear gain factor. Let L be Digital level in dBFS and A the corresponding ideal peak amplitude relative to a full-scale amplitude of 1. The sine and square RMS expressions below describe ideal complete cycles before fades.

Both A and R are dimensionless. A 20 dB reduction divides amplitude by ten; it does not describe a tenfold change in perceived loudness. The square RMS expression assumes an ideal waveform alternating between equal positive and negative levels. Bandlimiting changes that shape, and the finite fades change the whole-record RMS. For a sweep, the displayed RMS remains the ideal sine-cycle reference rather than an RMS calculation over its rendered samples.

Frequency trajectories

Let f₀ be Frequency, f₁ be End frequency, T be Duration in seconds, and t be elapsed time from 0 to T. Both frequencies are in hertz. A steady tone keeps f₀ throughout; the two sweep curves use these equations.

The linear midpoint is the arithmetic mean of the endpoints. The logarithmic midpoint is their geometric mean, the square root of their product. Reversing the endpoints reverses the sweep, and equal endpoints produce constant frequency. Positive frequency ratios also explain equal octave timing: each doubling occupies the same interval during a logarithmic sweep.

Amplitude envelope and continuous phase

A fade changes gain gradually to reduce abrupt transitions that can produce clicks. Let F be the fade time in seconds, obtained by dividing the millisecond value by 1,000. The envelope e(t) rises from 0 to 1, stays at 1, then falls to 0 within the total duration T.

For an active channel, the ideal peak envelope is A multiplied by e(t). The three interior checkpoints occur after the starting fade and before the ending fade, because duration is at least one second and each fade is at most 0.1 seconds. Their peaks therefore equal A. An inactive channel stays at zero.

A sine sweep accumulates phase continuously as frequency changes. In the ideal expression below, x(t) is signal amplitude and u is time within the integral. Integrating frequency counts the accumulated cycles; multiplying by 2π converts them to an angle.

The selected fades apply to scheduled start and completion. Manual Stop / Esc and setting changes normally use a separate 20 ms stop ramp. Page hiding or departure closes playback immediately, so the selected fade is not a guarantee for every interruption.

Sampling, harmonics and precision

Sample rate counts digital sample frames per second. Its half-rate is the Nyquist frequency, above which frequency content must be suppressed to avoid aliasing into lower frequencies. At 48 kHz, Nyquist is 24 kHz. Live playback rejects a starting or ending frequency at or above half its actual sample rate.

A square wave's odd harmonics extend above its fundamental, so a valid fundamental does not imply that every harmonic can be reproduced. Higher fundamentals leave room for fewer harmonics. Browser bandlimiting and waveform normalization mean that the ideal square preview and its RMS are not exact descriptions of the rendered square samples.

For stereo 16-bit PCM, each frame contains two two-byte samples. At the fixed 48 kHz WAV rate, the frame count N and file size B in bytes follow directly from duration T.

The 44 bytes account for the WAV header. Samples are rounded to signed 16-bit integers, adding quantization to the difference between an ideal mathematical wave and a stored file. The file's last sample occurs just before the mathematical endpoint; a zero endpoint in the checkpoint table does not promise an exactly zero final stored sample.

Computed values are displayed with up to four decimal places, while calculations and numeric CSV data retain their underlying precision. A displayed peak such as 0.0316 is rounded for reading; the signal is not synthesized from that shortened number.

Supported numerical limits

All listed bounds are inclusive. They define accepted signal settings, not acoustic safety thresholds.

Limitations and Privacy:

Tone generation and WAV rendering happen in the browser without microphone input. A guided speaker check organizes listening observations; a configurable tone generator defines a signal to repeat or reuse.

• Both channels receive the same signal at the same gain and phase. Choosing both does not halve the per-channel amplitude. Operating-system mono settings, balance, processing and connected hardware can change the final routing or sound. Check those settings before drawing conclusions from left-only or right-only playback.
• For repeat comparisons, keep device volume, audio processing, connected equipment and listening position consistent.
• Playback and WAV creation depend on browser audio support. If playback reports a Nyquist limit, lower the relevant frequency. If audio support is unavailable, use a browser with the required audio capabilities.
• Charts may need a network connection to load, and links that restore a configuration can include its signal settings.

Worked Examples:

These examples calculate the intended digital signal. They are not measurements of a speaker, headphone or listener.

References:

• Web Audio API 1.1, World Wide Web Consortium, Working Draft, 22 September 2026.
• Chirp frequency definitions, SciPy Manual.
• Deafness and hearing loss: Safe listening, World Health Organization, 6 March 2026.
• Envelopes, Carnegie Mellon University, Intro to Computer Music.
• Intervals, Exponents, Logarithms, Johns Hopkins University, Mathematics of Music.
• How to read and write WAV files with SciPy, Simplified Guide.