Live browser instrument · No account
Raise a hand, pinch to play, and move through the air to shape each note. This free gesture synthesizer gives each hand its own voice, so you can build melodies, intervals, and slow-moving drones with nothing more than a webcam.
Loading the gesture engine…
01 / HOW IT WORKS
Gesture Synth is deliberately closer to picking up an instrument than configuring music software. There is no account, installation, MIDI setup, or empty project to manage.
Press Start Camera and allow access when your browser asks. Audio begins from the same click because modern browsers require a deliberate gesture before sound can play. If a camera is not available, choose Mouse Mode instead.
Bring your thumb and index fingertip together. The pinch acts like pressing a key: close the gap to start a note and open your hand to release it. This clear gate keeps accidental movement from becoming a wall of sound.
Travel left and right to select notes from the current scale. Move upward for a brighter tone and downward for something warmer. Bring in your second hand for harmony, a bass interval, or a drifting two-note texture.
02 / GESTURE MAP
The interaction is intentionally small enough to learn by feel. Horizontal position chooses pitch. Vertical position controls filter brightness. A pinch controls note on and off. The rules stay the same for both hands, so the second voice feels like adding another key rather than learning another instrument.
03 / MAKE IT YOURS
Start with a major pentatonic scale for open, forgiving melodies. Switch to major for familiar seven-note movement or minor for a darker center. Changing the root moves the whole note field without changing the physical gesture you already learned.
Sine is soft and almost weightless. Triangle adds a little definition. Sawtooth makes the filter movement obvious, while square produces a hollow, game-like pulse. The same envelope keeps every voice responsive to quick pinches and long holds.
Mouse Mode uses the same pitch field and audio engine. Move the pointer to preview a position, hold to play, and drag to bend the musical phrase through scale steps. It is also the fastest way to understand the mapping before turning on a webcam.
04 / CAMERA PRIVACY
Camera access can feel unusually personal, so the boundary should be unusually clear. Gesture Synth requests a live stream only after you press Start Camera. MediaPipe Hand Landmarker processes those frames inside your browser and returns a small set of landmark coordinates. The musical mapping reads those coordinates; it does not need to save the image behind them.
There is no account system, upload endpoint, cloud recording, or hidden camera archive in this release. If you press Record, the browser creates a performance clip locally for preview, download, or system sharing; Gesture Synth does not upload that clip. Closing the page stops the tracks and clears the temporary clip. Pressing Stop Camera stops the camera without leaving the page. You can also deny camera permission and use Mouse Mode with the complete sound engine. Your browser may remember the permission choice, but you remain in control: camera access can be revoked from the site settings at any time. Gesture Synth does not request access to your photo library, files, contacts, location, or microphone.
05 / FAQ
Gesture Synth is a free browser instrument. It follows your hands through a webcam, converts stable landmark positions into musical controls, and plays sound through the Web Audio API. It is a synthesizer rather than a video filter: the goal is to make movement playable, repeatable, and expressive.
No. Hand landmark detection runs locally in your browser. Gesture Synth does not upload, store, or transmit your camera frames. The page receives coordinates such as fingertip position and pinch distance, then uses those numbers to control sound.
Yes. Choose Mouse Mode. Horizontal movement controls pitch, vertical movement controls brightness, and pressing the pointer gates the note. Mouse Mode uses the same scales, waveforms, and synthesis engine as the camera experience.
A current desktop version of Chrome or Edge offers the most consistent combination of webcam access, WebAssembly acceleration, and low-latency Web Audio. Safari and Firefox support many of the required APIs, but camera and audio behavior can vary between versions and devices.
Browsers block pages from playing sound or opening a camera without a deliberate user action. Pressing Start Camera both resumes the audio context and opens the permission prompt. Gesture Synth never attempts to activate either capability on page load.
The interface is responsive, but this first release is tuned for laptops and desktops. Mobile devices differ widely in camera orientation, sustained hand-tracking performance, heat, and audio latency. For the most reliable session, use a desktop browser with a front-facing webcam and even lighting.
Face the camera with light on your hands rather than behind them. Keep your full hand visible, leave some space around your fingers, and avoid a background that closely matches your skin tone. Slow, intentional movement is easier to track and usually sounds more musical than fast waving.
Yes. Start Camera or Mouse Mode, then press Record 15s in the instrument status bar. The browser creates a short performance clip with the synthesized audio, shows a local preview, and lets you download it or open the system share sheet when supported. The clip is not uploaded to Gesture Synth, and recording never requests microphone access.
READY SIGNAL