Ultrasonic Tone Generator

1. Check your hard ceiling first. The banner at the top of the tool reads your browser’s real audio sample rate and shows the Nyquist limit — the highest frequency it can possibly produce. You cannot set the slider or the number box above that value.
2. Pick a mode. Continuous lets you dial any frequency with the slider or type an exact value in hertz. Stepped walks the round checkpoints (15, 16, 17, 17.4, 18, 19, 20… up to your Nyquist). Log sweep glides from 15 kHz to the ceiling over about ten seconds.
3. Start the volume low and raise it gently. High tones fatigue ears without sounding loud, so never crank a tone you cannot hear.
4. Press Play, then use “Can you hear this?” In stepped mode, answer Yes/No at each step and click Higher to climb until the tone disappears.
5. Read the result honestly. Where a tone vanishes is set by your ears and your speaker/headphone together — the tool cannot tell which one ran out first.

This tool drives the Web Audio API’s oscillator at a frequency you choose and routes it through a gain stage to your speakers or headphones. You can pick a sine or a square waveform, but note that the oscillator is band-limited: it only synthesises harmonics below Nyquist. A square wave’s lowest extra harmonic sits at three times the fundamental, which for any tone in this generator’s 15 kHz-and-up range (45 kHz and higher) is already above the Nyquist ceiling, so it is discarded. In practice that means the square option here produces essentially the same pure tone as sine — the harmonics that would normally distinguish a square wave simply cannot exist at these frequencies. Because digital audio is sampled, there is a hard physical ceiling: by the Nyquist–Shannon sampling theorem, a system sampling at R samples per second can only represent frequencies below R/2. Browsers default to a 44,100 Hz sample rate (the CD rate), giving a Nyquist limit of 22,050 Hz; many devices run at 48,000 Hz, giving 24,000 Hz. The tool reads your actual AudioContext.sampleRate at runtime and caps both the slider and the number box at exactly half of it, so it will never let you request a tone it cannot generate. That is why “true ultrasound” — the dog-whistle and bat-call territory above ~22 kHz — is simply off the table for any normal web page.

The second limit is your transducer. Most consumer speakers and headphones are designed to cover roughly 20 Hz to 20 kHz, and many conventional tweeters begin rolling off well before 20 kHz; reaching cleanly above that usually needs a dedicated super-tweeter. Phone speakers, laptop speakers and Bluetooth links are typically worse and attenuate everything above ~16 kHz. So the generator can faithfully send a 21 kHz electrical signal to a device that physically cannot turn it into sound — you would hear nothing, and nothing would be wrong with the tool. Gain is ramped up and down with the Web Audio scheduler to avoid clicks, the default level is deliberately low, and we cap the peak gain because steady high tones can fatigue your ears without ever sounding loud.

Two honest cautions. First, “I can’t hear it” does not isolate the cause: a missing tone could be your hearing (the high end drops with age — presbycusis) or your hardware rolling off, and a pure generator with no microphone cannot separate the two. Second, because there is no microphone this tool measures nothing — it cannot report dB SPL, distortion, or a response curve. For real numbers you need a calibrated measurement microphone and software such as REW, and for genuine ultrasound capture you need an ultrasonic microphone or bat detector. Use this generator to produce high tones for listening tests and experiments, and reach for instruments when you need measurements.