Illustration by Angela Torchio
- npm release and Web Editor sync: CONTRIBUTING.md
- Web Editor batch workflow (CORS server, catalog updates): docs/web-editor/
- Migration log of sketch URLs: webeditorLinks.md
P5.js on mobile provides unique opportunities and challenges. The main P5 framework does an excellent job of making it easy to read data from various phone inputs and sensors, however it doesn't deal with the realities of contemporary browser's built in gestures and security protocols. That's where this library comes in:
- Simplifies accessing phone hardware from the browser (accelerometers, gyroscopes, microphone, vibration motor)
- Simplifies disabling default phone gestures (Zoom, refresh, back, etc)
- Simplifies enabling audio output
- Simplifies using an on-screen console to display errors and debug info
p5-phone ships two optional agent skills that teach AI coding assistants (Claude Code, Cursor, ZCode, OpenCode, GitHub Copilot, etc.) how the library and p5.js 2.x work, so generated sketches use the right APIs and contemporary patterns.
- p5-phone — knowledge of how the library works, implementation patterns, and commands to simplify creating applications built with this library.
- p5js-2x — while the main library is optimized for p5 2.x, many models are trained on p5 1.x methods. This skill bridges the transition by pointing development to contemporary methods.
The skill files live in .agents/skills/, .claude/skills/, .opencode/skills/, and .github/skills/ (kept byte-identical across all four). See the full guide for installing them for your particular workflow.
This library simplifies access to the following p5.js mobile sensor and audio commands:
Touch/Pointer Events:
mousePressed()- Called when a press/touch begins (works for both mouse and touch in p5.js 1.x and 2.0)mouseReleased()- Called when a press/touch ends (works for both mouse and touch in p5.js 1.x and 2.0)mouseIsPressed-truewhile a press/touch is down. It turnsfalseas soon as any one finger lifts, so multi-finger sketches should testtouches.length > 0insteadtouches- Every finger on the screen, each withx,yandidtouchStarted()- Called when a touch begins (p5.js 1.x only)touchEnded()- Called when a touch ends (p5.js 1.x only)
Device Motion & Orientation:
rotationX- Device tilt forward/backwardrotationY- Device tilt left/rightrotationZ- Device rotation around screenaccelerationX- Acceleration left/rightaccelerationY- Acceleration up/downaccelerationZ- Acceleration forward/backdeviceShaken()- Shake detection eventdeviceMoved()- Movement detection eventsetShakeThreshold()- Set shake detection sensitivitysetMoveThreshold()- Set movement detection sensitivity
rotationX/Y/Z follow angleMode(): they are radians unless the sketch calls angleMode(DEGREES).
Audio Input (requires p5.sound):
p5.AudioIn()- Audio input objectp5.Amplitude()+amplitude.getLevel()- Current audio input level (p5.sound 0.3.x has nomic.getLevel(); see Microphone Activation)
- iOS 13+ (Safari)
- Android 7+ (Chrome)
- Chrome 80+
- Safari 13+
- Firefox 75+
p5-phone supports both p5.js 1.x and p5.js 2.0+.
| Feature | p5.js 1.x | p5.js 2.0+ |
|---|---|---|
| Permission UI (Tap/Button/Canvas/Banner/Minimal/Custom) | ✅ | ✅ |
| Motion sensors (rotationX/Y/Z, accelerationX/Y/Z) | ✅ | ✅ |
| Microphone / Speech / Sound | ✅ | ✅ |
| Camera (PhoneCamera) | ✅ | ✅ |
| Vibration | ✅ | ✅ |
| NFC Tag Reading (Android only) | ✅ | ✅ |
| GPS / Geolocation (iOS + Android) | ✅ | ✅ |
| Bluetooth Low Energy (Web Bluetooth) | ✅ | ✅ |
| Share (multi-user rooms) | ✅ | ✅ |
| Desktop QR helper | ✅ | ✅ |
| Debug console | ✅ | ✅ |
| lockGestures() | ✅ | ✅ |
touchStarted() / touchEnded() |
✅ | ❌ Use mousePressed() / mouseReleased() |
p5.registerAddon() |
❌ | ✅ (auto-detected) |
Key change in p5.js 2.0: Touch-specific callbacks (touchStarted, touchMoved, touchEnded) are no longer dispatched. The unified Pointer API routes all input (mouse + touch) through mousePressed(), mouseDragged(), and mouseReleased(). These mouse callbacks work in both p5.js 1.x and 2.0, so use them for forward-compatible code.
Cancelled touches: p5.js 2.0 to 2.3.0 (2.2.3 included) have no pointercancel handler, so on their own they keep a touch the phone cancels (a system swipe, too many fingers) in touches until the page reloads. p5-phone releases that touch as if the finger lifted: it leaves touches, mouseIsPressed turns false, and mouseReleased(e) runs with e.type === 'pointercancel'. p5.js 2.3.1 and later release a cancelled touch themselves, without calling mouseReleased(), and p5-phone leaves it to them.
p5-phone automatically detects the p5.js version and adjusts its internal touch override behavior accordingly. No configuration needed.
- Link for Interactive Examples
- Maintainers
- Browser Compatibility
- p5.js Version Compatibility
- CDN (Recommended)
- Basic Setup
- API Reference
- Core Functions
- Status Variables
- Multiple Inputs and Outputs
- Motion Sensor Activation
- Microphone Activation
- Sound Output Activation
- Vibration Motor (Android Only)
- Torch / Flashlight (iOS + Android)
- NFC Tag Reading (Android Only)
- GPS / Geolocation (iOS + Android)
- Bluetooth Low Energy (Web Bluetooth)
- Share (multi-user shared state)
- Speech Recognition Activation
- PhoneCamera (ML5 Integration)
- Utilities
- Permission UI Styles
- Troubleshooting / FAQ
<!-- Minified version (recommended) -->
<script src="https://cdn.jsdelivr.net/npm/p5-phone@1.15.4/dist/p5-phone.min.js"></script>
<!-- Development version (larger, with comments) -->
<!-- <script src="https://cdn.jsdelivr.net/npm/p5-phone@1.15.4/dist/p5-phone.js"></script> --><!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Mobile p5.js App</title>
<!-- Basic CSS to remove browser defaults and align canvas -->
<style>
body {
margin: 0;
padding: 0;
overflow: hidden;
}
</style>
<!-- Load p5.js library -->
<script src="https://cdn.jsdelivr.net/npm/p5@2.2.3/lib/p5.js"></script>
<script src="https://cdn.jsdelivr.net/npm/p5.js-compatibility@0.2.0/src/preload.js"></script>
<!-- Load p5-phone library -->
<script src="https://cdn.jsdelivr.net/npm/p5-phone@1.15.4/dist/p5-phone.min.js"></script>
</head>
<body>
<!-- Load the p5.js sketch -->
<script src="sketch.js"></script>
</body>
</html>let mic;
let amplitude;
let mySound;
function preload() {
// Load sound file if needed
// mySound = loadSound('assets/sound.mp3');
}
function setup() {
// Show debug panel FIRST to catch setup errors
showDebug();
createCanvas(windowWidth, windowHeight);
// Lock mobile gestures to prevent browser interference
lockGestures();
// Tilt in degrees (rotationX/Y/Z follow angleMode, and p5 uses radians unless told)
angleMode(DEGREES);
// Set up the microphone before the tap
mic = new p5.AudioIn();
amplitude = new p5.Amplitude();
mic.disconnect(); // keep the live mic out of the speakers
mic.connect(amplitude); // read the level with amplitude.getLevel()
// One tap asks for motion, the microphone and sound output together.
// Use a single enable* call: a second one would replace this tap screen.
enablePermissionsTap(['gyro', 'mic', 'sound'], 'Tap to start');
// For one feature on its own, the single versions work the same way:
// enableGyroTap('Tap to enable motion sensors');
}
function draw() {
background(220);
// Always check status before using hardware features
if (window.sensorsEnabled) {
// Use device rotation and acceleration
fill(255, 0, 0);
circle(width/2 + rotationY * 5, height/2 + rotationX * 5, 50);
}
if (window.micOpen) {
// Use microphone input (micOpen is true only while the mic is really live)
let level = amplitude.getLevel();
fill(0, 255, 0);
rect(10, 10, level * 200, 20);
}
if (window.soundEnabled) {
// Safe to play sounds
// mySound.play();
}
}
// Prevent default touch behavior (optional but recommended)
// Use mousePressed/mouseReleased — works in both p5.js 1.x and 2.0
function mousePressed() {
return false;
}
function mouseReleased() {
return false;
}Most sketches need more than one thing, like motion and the microphone. List them all in one enablePermissionsTap() call, and one tap asks for everything. Don't call two enable* functions in a row: each one removes the tap screen of the one before, so only the last one would ask.
let mic;
let amplitude;
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
angleMode(DEGREES);
mic = new p5.AudioIn(); // the microphone needs p5.sound
amplitude = new p5.Amplitude();
mic.disconnect(); // keep the mic out of the speakers
mic.connect(amplitude);
// one tap asks for both
enablePermissionsTap(['gyro', 'mic'], 'Tap to start');
}
function draw() {
background(20);
// check each one on its own
if (window.sensorsEnabled) {
circle(width / 2 + rotationY * 3, height / 2 + rotationX * 3, 60);
}
if (window.micOpen) {
rect(0, height - 20, amplitude.getLevel() * width, 20);
}
}The list can hold any of gyro (motion), mic, sound, speech, vibration, torch, nfc, geo and camera. Each one sets its own flag: gyro sets window.sensorsEnabled, mic sets window.micOpen, torch sets window.torchEnabled, and so on. See Multiple Inputs and Outputs for the full list, the other styles, and adding Bluetooth or Share.
// Essential mobile setup
lockGestures() // Prevent browser gestures (call in setup())
// Motion sensor activation (enableSensor* is another name for the same functions)
enableGyroTap(message) // Tap anywhere to enable sensors
enableGyroButton(text) // Button-based sensor activation
// Microphone activation
enableMicTap(message) // Tap anywhere to enable microphone
enableMicButton(text) // Button-based microphone activation
// Sound output activation (no microphone input)
enableSoundTap(message) // Tap anywhere to enable sound playback
enableSoundButton(text) // Button-based sound activation
// Speech recognition activation (Web Speech API)
enableSpeechTap(message) // Tap anywhere to enable speech
enableSpeechButton(text) // Button-based speech activation
// Combined activation (motion + microphone)
enableAllTap(message) // Tap anywhere to enable both
enableAllButton(text) // Button-based combined activation
// Any combination of hardware permissions
enablePermissionsTap(['gyro', 'mic', 'camera'], message)
enablePermissionsButton(['torch', 'vibration'], text, statusText)
enablePermissionsCanvas(['camera', 'mic'], message)
enablePermissionsBanner(['gyro', 'nfc'], message, position)
enablePermissionsOn('#start-button', ['camera', 'mic'])
// Also: enablePermissionsMinimal(list, message | options)
// Also available as enableHardwareTap/Button/Canvas/Banner/Minimal/On
// Vibration motor (Android only)
enableVibrationTap(message) // Tap anywhere to enable vibration
enableVibrationButton(text) // Button-based vibration activation
vibrate(pattern) // Trigger vibration (duration or pattern array)
stopVibration() // Stop any ongoing vibration
// Torch / flashlight (iPhone iOS 17.4+ and Android Chrome, HTTPS required)
enableTorchTap(message) // Tap anywhere to enable flashlight control
enableTorchButton(text) // Button-based flashlight activation
torchOn() // Turn flashlight on
torchOff() // Turn flashlight off
toggleTorch() // Toggle flashlight state
setTorch(true) // Set flashlight to a boolean state
stopTorch() // Turn off and release the camera stream
// Flashlight aliases also work: enableFlashlightTap(), flashlightOn(), toggleFlashlight(), stopFlashlight(), etc.
// Camera (ML5 integration)
createPhoneCamera(active, mirror, mode) // Create camera instance
enableCameraTap(message) // Tap to enable camera
enableCameraButton(text) // Button-based camera activation
// GPS / geolocation (iOS Safari + Android Chrome, HTTPS required)
enableGeoTap(message) // Tap anywhere to enable GPS
enableGeoButton(text) // Button-based GPS activation
setGeoOptions({ enableHighAccuracy, timeout, maximumAge }) // Tune before enableGeo* (default: coarse)
getGeoPosition() // Return last position synchronously (or null)
geoDistance(lat1, lon1, lat2, lon2, units) // Haversine distance: 'm' (default), 'km', or 'mi'
geoInPolygon(polygon, point) // Point-in-geofence test (ray casting)
stopGeo() // Stop the GPS watch and release the subscription
// Bluetooth LE and multi-user rooms: see their sections below
// bleSetup(), enableBleTap(), bleConnect(), bleWrite(), bleRead(), bleDisconnect()
// shareSetup(), enableShareTap(), shared / me / guests, shareEmit(), shareDisconnect()
// Desktop QR helper (no-op on phones) and device detection (v1.14.0)
showDesktopQr(options) // Floating QR of this page on desktop; scan it to open the sketch on a phone
setQrUrl(url) // Point the QR at a different URL
hideDesktopQr() // Remove the QR panel
window.isMobile // true on phones and tablets (best effort)
window.isDesktop // true otherwise
// --- Alternative Permission UI Styles (v1.7.0) ---
// Canvas-first-touch — permissions fire on first canvas interaction
enableGyroCanvas(message) // Also: enableMicCanvas, enableSoundCanvas,
// enableSpeechCanvas, enableVibrationCanvas, enableTorchCanvas,
// enableAllCanvas, enableNfcCanvas, enableGeoCanvas,
// enableCameraCanvas, enableBleCanvas(opts), enableShareCanvas(opts)
// Banner — slim notification bar at top or bottom of screen
enableGyroBanner(message, position) // position: 'top' or 'bottom'
// Also: enableMicBanner, enableSoundBanner,
// enableSpeechBanner, enableVibrationBanner, enableTorchBanner,
// enableAllBanner, enableNfcBanner, enableGeoBanner,
// enableCameraBanner, enableBleBanner(opts), enableShareBanner(opts)
// Custom element binding — attach to your own DOM element
enableGyroOn(selector) // e.g., enableGyroOn('#my-button')
// Also: enableMicOn, enableSoundOn, enableSpeechOn,
// enableVibrationOn, enableTorchOn, enableAllOn, enableNfcOn, enableGeoOn,
// enableCameraOn, enableBleOn, enableShareOn
// Minimal — bare semi-transparent overlay with a pulsing icon (v1.14.0)
enableGyroMinimal(message) // or enableGyroMinimal({ color, opacity, icon, iconColor, iconSize, message })
// Also: enableMicMinimal, enableSoundMinimal, enableSpeechMinimal,
// enableVibrationMinimal, enableTorchMinimal, enableNfcMinimal,
// enableGeoMinimal, enableAllMinimal, enableCameraMinimal,
// enableBleMinimal(opts), enableShareMinimal(opts)
// Status variables (check these in your code)
window.sensorsEnabled // Boolean: true when motion sensors are active
window.micEnabled // Boolean: true once the microphone request has run (NOT proof of access)
window.micOpen // Boolean: true only while the microphone stream is really live
window.soundEnabled // Boolean: true when sound output is active
window.speechEnabled // Boolean: true when speech recognition is active
window.vibrationEnabled // Boolean: true when vibration is available (Android only)
window.torchEnabled // Boolean: true when torch camera stream is active
window.torchSupported // Boolean: true when torch support is reported
window.torchActive // Boolean: true when flashlight is currently on
window.nfcEnabled // Boolean: true when NFC scanning is active (Android only)
window.geoEnabled // Boolean: true when GPS watch is active (iOS + Android)
window.cameraEnabled // Boolean: true after camera startup succeeds
// Debug system
showDebug() // Show on-screen debug panel with automatic error catching
hideDebug() // Hide debug panel
toggleDebug() // Toggle panel visibility
debug(...args) // Console.log with on-screen display and timestamps
debugError(...args) // Display errors with red styling
debugWarn(...args) // Display warnings with yellow styling
debug.clear() // Clear debug messagesp5.js Namespace Support: All functions are also available as p5.prototype methods:
// You can use either syntax:
lockGestures(); // Global function (recommended)
this.lockGestures(); // p5.js instance method
// Both approaches work identically
enableGyroTap('Tap to start');
this.enableGyroTap('Tap to start');Purpose: Check whether permissions have been granted and sensors are active.
Variables:
window.sensorsEnabled- Boolean indicating if motion sensors are activewindow.micEnabled- Boolean indicating the microphone request has run. It istrueeven if the person denies access or no input device existswindow.micOpen- Boolean,trueonly while the microphone stream is really live. Use this to tell a refused microphone from silencewindow.soundEnabled- Boolean indicating if sound output is activewindow.speechEnabled- Boolean indicating if speech recognition is activewindow.vibrationEnabled- Boolean indicating if vibration is available (Android only)window.torchEnabled- Boolean indicating if the torch control stream is active (iOS + Android)window.torchSupported- Boolean indicating if the active camera track reports torch support (falsewith no rear flash or an older browser)window.torchActive- Boolean indicating if the flashlight is currently onwindow.torchError- Last torch error message, if anywindow.nfcEnabled- Boolean indicating if NFC scanning is active (Android only)window.geoEnabled- Boolean indicating if the GPS watch is active (iOS + Android)window.geoStatus- String describing GPS status (idle/requesting-permission/active/permission-denied/unsupported/secure-context-required/error/stopped)window.geoError- String containing the latest GPS error messagewindow.lastGeoPosition- Most recent normalized position{ latitude, longitude, accuracy, altitude, altitudeAccuracy, heading, speed, timestamp }window.cameraEnabled- Boolean indicating if camera startup has succeededwindow.lastNfcSerialNumber- Serial number string for the most recently read NFC tagwindow.lastNfcAlias- Alias string for the most recently read NFC tag, if one has been setwindow.shareConnected- Boolean indicating if the Share PartyServer room is connectedwindow.shareStatus- Share status (idle/connecting/connected/error/unsupported)window.shareIsHost- Boolean indicating if this client is the room hostshared/me/guests- Live multi-user objects (see Share section)window.gesturesLocked- Boolean,trueafterlockGestures()has installed its handlerswindow.isMobile/window.isDesktop- Best-effort device detection (see Desktop QR Helper and Device Detection)
Usage:
function draw() {
// Always check before using sensor data
if (window.sensorsEnabled) {
// Safe to use rotationX, rotationY, accelerationX, etc.
let tilt = rotationX;
}
if (window.micOpen) {
// Microphone is really delivering audio
let audioLevel = amplitude.getLevel();
}
if (window.soundEnabled) {
// Safe to play sounds
mySound.play();
}
if (window.vibrationEnabled) {
// Safe to use vibration (Android only)
vibrate(50);
}
if (window.nfcEnabled) {
// NFC scanning is active (Android only)
// Tag data arrives via nfcRead() callback
}
}
// You can also use them for conditional UI
function draw() {
if (window.sensorsEnabled) {
debug("Motion sensors enabled");
} else {
debug("Motion sensors not yet enabled");
}
}Purpose: Ask for several features from one tap: motion and sound, the microphone and the flashlight, or any other mix.
The idea: Browsers grant most phone hardware only from a tap, and one tap can ask for several things. Put everything the sketch needs in one list and pass it to a single enablePermissions* call. Each feature then sets its own status flag, so check them one by one. A phone can grant one feature and not another (an iPhone has no vibration or NFC), and the rest of the sketch keeps working.
Use one enable* call. Each enable* function removes the tap screen, button, banner or overlay of the one before, so two calls in a row leave only the last one, and only that one asks:
// ✗ only the microphone asks: enableMicTap() removed the motion tap screen
enableGyroTap('Tap for motion');
enableMicTap('Tap for the microphone');
// ✓ one tap asks for both
enablePermissionsTap(['gyro', 'mic'], 'Tap to start');Commands (every style takes the same list):
enablePermissionsTap(list, message)- a full-screen tap. The usual choice.enablePermissionsButton(list, buttonText, statusText)- a generated buttonenablePermissionsCanvas(list, message)- the first touch on the canvas, with no overlayenablePermissionsBanner(list, message, position)- a slide-in banner,'top'or'bottom'enablePermissionsMinimal(list, message | options)- a bare overlay with a pulsing iconenablePermissionsOn(selector, list)- your own HTML element. The selector comes first here.enableHardwareTap(list, message)and the rest ofenableHardware*- the same six functions under another nameenableAllTap(message)and the rest ofenableAll*- shorthand for['gyro', 'mic']
The list is an array, or a space- or comma-separated string, in any order:
| In the list | Turns on | Check this flag | Other names that work |
|---|---|---|---|
gyro |
motion and orientation | window.sensorsEnabled |
motion, sensors, sensor, orientation, gyroscope, accelerometer |
mic |
microphone | window.micOpen |
microphone, audioin |
sound |
sound output | window.soundEnabled |
audio, audiooutput, output |
speech |
speech recognition | window.speechEnabled |
voice, recognition |
vibration |
vibration (Android) | window.vibrationEnabled |
vibrate, haptic, haptics |
torch |
flashlight | window.torchEnabled |
flashlight, flash, light |
nfc |
NFC tags (Android) | window.nfcEnabled |
tag, tags |
geo |
GPS / location | window.geoEnabled |
gps, location, geolocation |
camera |
PhoneCamera |
window.cameraEnabled |
video, webcam |
userSetupComplete() runs once, after every request in the tap has finished.
A complete sketch is in the Quick Start: Several inputs and outputs from one tap.
Bluetooth and Share as well. They are not in the list, and a second enable* call in setup() would remove the first one's tap screen. Start them once the first tap has worked. The Bluetooth chooser needs a tap of its own, so show its button from userSetupComplete():
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
bleSetup({ characteristics: [{ name: 'brightness', type: 'uint8', write: true }] });
enablePermissionsTap(['gyro', 'sound'], 'Tap to start');
}
function userSetupComplete() {
// runs after the first tap. the Bluetooth chooser needs a tap of its own
enableBleButton({ label: 'Connect the lamp' });
}For Share, call shareSetup() in setup() and shareConnect() in userSetupComplete(). Share needs no second tap.
Purpose: Enable device motion and orientation sensors with user permission handling.
Commands:
enableGyroTap(message)- Tap anywhere on screen to enable sensorsenableGyroButton(text)- Creates a button with custom text to enable sensors
Usage:
// Tap-to-enable (recommended)
enableGyroTap('Tap to enable motion sensors');
// Button-based activation
enableGyroButton('Enable Motion');Available p5.js Variables (when window.sensorsEnabled is true):
| Variable | Description | Range/Units |
|---|---|---|
rotationX |
Device tilt forward/backward | -180° to 180° (with angleMode(DEGREES)) |
rotationY |
Device tilt left/right | -90° to 90° (with angleMode(DEGREES)) |
rotationZ |
Device rotation around screen | 0° to 360° (with angleMode(DEGREES)) |
accelerationX |
Acceleration left/right | m/s² |
accelerationY |
Acceleration up/down | m/s² |
accelerationZ |
Acceleration forward/back | m/s² |
deviceShaken |
Shake detection event | true when shaken |
deviceMoved |
Movement detection event | true when moved |
Important: All motion sensor variables, including deviceShaken and deviceMoved, are only available when window.sensorsEnabled is true. Always check this status before using any motion data.
Degrees or radians: rotationX/Y/Z follow angleMode(), and p5.js uses radians unless told otherwise. Call angleMode(DEGREES) in setup() for the degree ranges above; in radians, rotationY * 3 moves a circle a few pixels.
On a laptop: a computer with no tilt sensor still turns window.sensorsEnabled true after the tap, and rotationX/Y/Z stay 0. The browser sends one empty reading there (every angle null); p5-phone drops it before p5 sees it, so map(rotationX, …) never gets null.
Example:
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
angleMode(DEGREES); // rotationX/Y/Z in degrees
enableGyroTap('Tap to enable motion sensors');
}
function draw() {
// CRITICAL: Always check window.sensorsEnabled first
if (window.sensorsEnabled) {
// Tilt-controlled circle
let x = width/2 + rotationY * 3;
let y = height/2 + rotationX * 3;
circle(x, y, 50);
// Shake detection - only works when sensors are enabled
if (deviceShaken) {
background(random(255), random(255), random(255));
}
// Movement detection - also requires sensors to be enabled
if (deviceMoved) {
fill(255, 0, 0);
}
} else {
// Show fallback when sensors not enabled
text('Tap to enable motion sensors', 20, 20);
}
}Purpose: Enable device microphone with user permission handling for audio-reactive applications.
Important: Microphone examples require the p5.sound library. Add this script tag to your HTML:
<script src="https://cdn.jsdelivr.net/npm/p5.sound@0.3.0/dist/p5.sound.min.js"></script>Commands:
enableMicTap(message)- Tap anywhere on screen to enable microphoneenableMicButton(text)- Creates a button with custom text to enable microphone
Usage:
// Tap-to-enable (recommended)
enableMicTap('Tap to enable microphone');
// Button-based activation
enableMicButton('Enable Audio');Available p5.js Variables:
| Variable | Description | Range |
|---|---|---|
p5.AudioIn() |
Audio input object (stored in mic) |
Object |
p5.Amplitude() |
Level analyzer the mic is routed into | Object |
amplitude.getLevel() |
Current audio input level | 0.0 to 1.0 |
window.micEnabled |
The microphone request has run | Boolean |
window.micOpen |
The microphone stream is really live | Boolean |
Reading the level (p5.sound 0.3.x / p5.js 2.x): p5.AudioIn has no getLevel() in p5.sound 0.3.x. Route the mic into a p5.Amplitude and read that instead. Call mic.disconnect() first: every p5.sound 0.3.x node is wired to the speakers by default, so without it the live microphone plays straight out of the phone (feedback).
micEnabled vs micOpen: window.micEnabled becomes true as soon as p5-phone has asked for the microphone, even if the person then taps "Don't Allow" or the device has no input. p5.sound 0.3.x swallows that failure (it only logs mic not open), so a refused microphone looks exactly like silence. window.micOpen is true only while the stream is actually live, and drops back to false if it ends. It stays false while the browser prompt is still showing.
Example:
let mic;
let amplitude;
function setup() {
createCanvas(windowWidth, windowHeight);
// Create a new p5.AudioIn() instance
mic = new p5.AudioIn();
// Route the mic into an analyzer instead of the speakers
amplitude = new p5.Amplitude();
mic.disconnect();
mic.connect(amplitude);
// Enable microphone with tap
enableMicTap();
}
function draw() {
if (window.micOpen) {
// Audio-reactive visualization
let level = amplitude.getLevel();
let size = map(level, 0, 1, 10, 200);
background(level * 255);
circle(width/2, height/2, size);
} else if (window.micEnabled) {
// Asked, but no audio: still prompting, denied, or no input device
background(40);
fill(255);
text('Microphone not available', 20, 40);
}
}Using the legacy p5.sound that ships with p5.js 1.x?
mic.getLevel()still works there, andwindow.micOpenworks the same way.
Purpose: Enable audio playback without requiring microphone input. Perfect for playing sounds, music, synthesizers, and audio effects in mobile browsers.
Important: Sound examples require the p5.sound library. Add this script tag to your HTML:
<script src="https://cdn.jsdelivr.net/npm/p5.sound@0.3.0/dist/p5.sound.min.js"></script>Other sound libraries: the tap starts whatever is loaded, not only p5.sound:
- Tone.js: the tap calls
Tone.start(). Load Tone.js before or after p5-phone. - Your own
AudioContext(for example for smplr): the tap resumes anynew AudioContext()made after p5-phone loads, so put your<script>after p5-phone's.
The tap starts audio before it asks for anything else, so enablePermissionsTap(['motion', 'sound']) and enableAllTap() work on iPhone, where the motion prompt would otherwise end the tap first.
After the first tap: phones pause audio when you leave the page or lock the screen. The next touch, click, or key starts it again, so you don't need userStartAudio() in mousePressed(). Audio your sketch paused on purpose (userStopAudio(), ctx.suspend()) stays paused until your sketch resumes it.
Commands:
enableSoundTap(message)- Tap anywhere on screen to enable sound playbackenableSoundButton(text)- Creates a button with custom text to enable sound
Usage:
// Tap-to-enable (recommended)
enableSoundTap('Tap to enable sound');
// Button-based activation
enableSoundButton('Enable Sound');When to use Sound vs. Microphone:
- Use
enableSoundfor: Playing audio files, synthesizers, oscillators, sound effects - Use
enableMicfor: Recording audio, audio-reactive visualizations, voice input - Note:
enableMicalso enables sound output, so you don't need both
Example:
let mySound;
function preload() {
// Load audio file
mySound = loadSound('assets/sound.mp3');
}
function setup() {
createCanvas(windowWidth, windowHeight);
// Enable sound playback with tap
enableSoundTap('Tap to enable sound');
}
function draw() {
background(220);
if (window.soundEnabled) {
text('Tap anywhere to play sound', 20, 20);
} else {
text('Waiting for sound activation...', 20, 20);
}
}
function mousePressed() {
// Check if sound is enabled before playing
if (window.soundEnabled && !mySound.isPlaying()) {
mySound.play();
}
}Purpose: Access the device's vibration motor for haptic feedback and tactile interactions.
- ✅ Android - Full support in Chrome and most Android browsers
- ❌ iOS - Not supported (Vibration API not available on iOS devices)
Important: The vibration feature will automatically detect if the device supports vibration. On iOS or unsupported devices, window.vibrationEnabled will be false and vibration calls will be safely ignored with console warnings.
Commands:
enableVibrationTap(message)- Tap anywhere on screen to enable vibrationenableVibrationButton(text)- Creates a button with custom text to enable vibrationvibrate(pattern)- Trigger vibration with a duration (ms) or pattern arraystopVibration()- Stop any ongoing vibration
Usage:
function setup() {
createCanvas(windowWidth, windowHeight);
// Enable vibration with tap (Android only)
enableVibrationTap('Tap to enable vibration');
// Or use a button
// enableVibrationButton('Enable Haptics');
}
function draw() {
background(220);
if (window.vibrationEnabled) {
text('Vibration ready! Tap anywhere', 20, 20);
} else {
text('Vibration not available', 20, 20);
}
}
function mousePressed() {
if (window.vibrationEnabled) {
// Simple vibration - 50ms pulse
vibrate(50);
}
}Vibration Patterns:
// Single vibration (duration in milliseconds)
vibrate(100); // Vibrate for 100ms
// Pattern: [vibrate, pause, vibrate, pause, ...]
vibrate([100, 50, 100]); // Short-short pattern
vibrate([200, 100, 200, 100, 200]); // Triple pulse
vibrate([50, 50, 50, 50, 500]); // Quick taps then long
// Stop any ongoing vibration
stopVibration();Common Use Cases:
// Haptic feedback for button presses
function mousePressed() {
if (window.vibrationEnabled) {
vibrate(20); // Quick tap feedback
}
}
// Touch zones with different haptic patterns
function mousePressed() {
if (window.vibrationEnabled) {
if (mouseX < width/2) {
vibrate(50); // Left side - short pulse
} else {
vibrate([50, 30, 50]); // Right side - double pulse
}
}
return false;
}
// Collision detection
function checkCollision() {
if (collision && window.vibrationEnabled) {
vibrate([100, 50, 100, 50, 200]); // Alert pattern
}
}
// Game events
function gameOver() {
if (window.vibrationEnabled) {
vibrate(500); // Long vibration for game over
}
}Best Practices:
- Use short vibrations (20-100ms) for subtle feedback
- Use patterns for more complex haptic responses
- Always check
window.vibrationEnabledbefore callingvibrate() - Don't overuse - vibration can quickly drain battery
- Test on Android devices as iOS doesn't support vibration
Purpose: Control the rear camera flashlight from a sketch. This is useful for light-based interactions, signaling, installations, and hardware feedback.
Platform Support:
- iOS Safari - Supported on iPhone with iOS 17.4 or later
- Android Chrome / Chromium browsers - Supported on devices that expose camera torch constraints
- Requires HTTPS, camera permission, and a phone with a rear flash
- Check
isTorchSupported()after enabling. It returnsfalseon phones with no rear flash, in older browsers, and on desktop.
Important: Browser flashlight control is provided through a rear camera MediaStreamTrack, not through a separate flashlight permission. enableTorch*() starts an internal rear camera stream and keeps it alive while you control the torch. Use stopTorch() when you are done to turn the flashlight off and release the camera.
Commands:
enableTorchTap(message)- Tap anywhere on screen to enable flashlight controlenableTorchButton(text)- Creates a button with custom text to enable flashlight controltorchOn()- Turn the flashlight ontorchOff()- Turn the flashlight off while keeping control availabletoggleTorch()- Toggle the flashlight statesetTorch(enabled)- Set the flashlight to a specific boolean statestopTorch()- Turn off and release the internal camera streamisTorchSupported()- Check whether the active camera track reports torch support
Aliases: All torch APIs also have flashlight aliases, such as enableFlashlightTap(), flashlightOn(), flashlightOff(), toggleFlashlight(), and setFlashlight().
Usage:
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
enableTorchTap('Tap to enable flashlight');
}
function draw() {
background(window.torchActive ? 255 : 20);
fill(window.torchActive ? 20 : 255);
text(window.torchActive ? 'on' : 'off', 30, 40);
}
async function mousePressed() {
if (window.torchEnabled) {
await toggleTorch();
}
return false;
}Combined Hardware:
// Flashlight plus vibration
enablePermissionsTap(['torch', 'vibration'], 'Tap to enable disco hardware');
// Motion sensors plus flashlight
enablePermissionsTap(['gyro', 'torch'], 'Tap to enable shake flashlight');Status Variables:
window.torchEnabled- Boolean indicating if the internal torch camera stream is activewindow.torchSupported- Boolean indicating if the active track reports torch supportwindow.torchActive- Boolean indicating if the flashlight is currently onwindow.torchError- Latest torch error message, if anywindow.torchCapability- Raw torch capability value reported by the browser
Best Practices:
- Use short, slow pulses if flashing the light
- Avoid rapid strobing
- Call
torchOff()before pausing a light effect - Call
stopTorch()when the sketch no longer needs flashlight control - Test on real phones (iPhone and Android) because browser and hardware support varies
Purpose: Read NFC (Near Field Communication) tags using the Web NFC API. Ideal for interactive installations, scavenger hunts, or any sketch that responds to physical NFC tags.
- ✅ Android - Chrome 89+ and Samsung Internet 15+
- ❌ iOS - Not supported (Web NFC API not available on iOS)
- ❌ Inside an iframe (the p5.js Web Editor, embeds) - Web NFC only works when the sketch is the whole page. There
window.nfcStatusis'unsupported'andwindow.nfcErrorsays to open the sketch on its own page, such as GitHub Pages. - Requires HTTPS — NFC is blocked on insecure origins
Important: The Web NFC API requires user activation (a tap or click) before scanning can begin — the same pattern used by all other p5-phone permission functions. On unsupported devices/browsers, window.nfcEnabled will be false and calls will be safely ignored with console warnings. Compatible with widely available NFC Type 2 tags — including NTAG213, NTAG215, and NTAG216 — as well as any NDEF-formatted tag.
Commands:
enableNfcTap(message)- Tap anywhere on screen to enable NFC scanningenableNfcButton(text)- Creates a button with custom text to enable NFCstopNfc()- Stop NFC scanningsetNfcTagAlias(serialNumber, alias)- Give a tag ID a human-friendly namegetNfcTagAlias(serialNumber)- Get a saved alias for a tag IDisNfcTag(aliasOrSerialNumber)- Check whether the most recently read tag matches an alias or ID
Status Variables:
window.nfcEnabled- Boolean indicating if NFC scanning is activewindow.nfcStatus- String describing NFC startup/read statuswindow.nfcError- String containing the latest NFC startup/read errorwindow.nfcTagAliases- Object mapping tag IDs to aliaseswindow.lastNfcMessage- Object containing the most recently read tag's datawindow.lastNfcSerialNumber- Serial number string of the most recently read tagwindow.lastNfcAlias- Alias string for the most recently read tag, if one has been set
Two-step workflow:
- Open the NFC identifier example, scan each physical tag, type an alias, then download the tag list.
- Paste the generated
setNfcTagAlias()lines into your sketch and useisNfcTag()in conditionals.
User Callback:
Define an nfcRead() function in your sketch to receive tag data when a tag is scanned:
function nfcRead(message, serialNumber) {
// message.serialNumber — tag serial number
// message.alias — saved alias, or '' if unnamed
// message.records — array of NDEF records, each with:
// .recordType — 'text', 'url', 'mime', etc.
// .data — decoded content (string for text/url, object for JSON, raw for others)
// .mediaType — MIME type (for 'mime' records)
// .id — record id (if present)
// .raw — original DataView
}Usage:
let tagText = 'No tag scanned yet';
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
// Paste IDs from the NFC identifier example.
setNfcTagAlias('04:85:2a:1b:9f:61:80', 'paintbrush');
setNfcTagAlias('04:42:18:3c:9f:61:80', 'desk');
enableNfcTap('Tap to enable NFC');
}
function draw() {
background(220);
textAlign(CENTER, CENTER);
textSize(20);
if (window.nfcEnabled) {
if (isNfcTag('paintbrush')) {
background(120, 220, 160);
text('Paintbrush tag read', width / 2, height / 2);
} else if (isNfcTag('desk')) {
background(140, 180, 240);
text('Desk tag read', width / 2, height / 2);
} else {
text(tagText, width / 2, height / 2);
text('Hold an NFC tag near your phone', width / 2, height / 2 + 40);
}
} else {
text('NFC not active', width / 2, height / 2);
}
}
function nfcRead(message, serialNumber) {
tagText = 'Tag: ' + (message.alias || serialNumber);
if (isNfcTag('paintbrush', serialNumber)) {
// This runs once when the paintbrush tag is scanned.
}
for (let record of message.records) {
if (record.recordType === 'text' || record.recordType === 'url') {
tagText += '\n' + record.data;
}
}
}Record Types:
NFC tags contain NDEF records. The most common types are:
| Record Type | record.data Contains |
Example |
|---|---|---|
text |
Decoded string | "Hello World" |
url |
Decoded URL string | "https://example.com" |
mime |
Decoded string or parsed JSON | { id: 42 } |
| other | Raw DataView |
Binary data |
Best Practices:
- Always check
window.nfcEnabledbefore relying on NFC features - Use the
nfcRead()callback for real-time tag processing - Use
setNfcTagAlias()andisNfcTag()when a sketch should respond to named physical objects - Use
window.lastNfcMessageindraw()for displaying the most recent tag - Test on Android devices with Chrome — NFC is not available on iOS or desktop browsers
- Tags must be NDEF-formatted to be read by the Web NFC API. Widely available NFC Type 2 tags work out of the box: NTAG213 (144 bytes), NTAG215 (504 bytes, also used for Amiibo), and NTAG216 (888 bytes)
Purpose: Read the device's geographic position using the browser Geolocation API (navigator.geolocation). Ideal for location-aware sketches, distance/wayfinding visualizations, geofencing, and any work that responds to where the phone is. Unlike NFC and vibration, GPS works on both iOS Safari and Android Chrome.
✅ Platform Support:
- ✅ iOS - Safari (iOS 13+), requires HTTPS and a user gesture to prompt
- ✅ Android - Chrome, requires HTTPS
- Requires HTTPS — geolocation is blocked on insecure origins (
http://), exceptlocalhost
Important: GPS permission must be requested from a user gesture (tap/click) — the same pattern used by all other p5-phone permission functions. The first position fix can take 5-30 seconds (cold start) as the GPS warms up; window.geoStatus will read 'requesting-permission' during this time. By default the library uses a coarse, battery-friendly fix (Wi-Fi/cell, ~50-100m). Call setGeoOptions({ enableHighAccuracy: true }) before enabling to opt into real GPS (~5-10m outdoors, slower, more battery).
Commands:
enableGeoTap(message)- Tap anywhere on screen to enable GPSenableGeoButton(text)- Creates a button with custom text to enable GPSstopGeo()- Stop the GPS watch and release the position subscriptionsetGeoOptions(opts)- Tune accuracy/timeout/maximumAge (call beforeenableGeo*)getGeoPosition()- Return the last known position synchronously, ornullgeoDistance(lat1, lon1, lat2, lon2, units)- Great-circle distance (Haversine);unitsis'm'(default),'km', or'mi'geoInPolygon(polygon, point)- Point-in-geofence test (ray casting);polygonis[{lat, lon}, ...],pointis{lat, lon}
Status Variables:
window.geoEnabled- Boolean indicating if the GPS watch is activewindow.geoStatus- String describing GPS status (idle/requesting-permission/active/permission-denied/unsupported/secure-context-required/error/stopped)window.geoError- String containing the latest GPS error messagewindow.lastGeoPosition- Most recent normalized position object
User Callbacks:
Define a geoRead(position) function in your sketch to receive position updates. Define onGeoError(error) to receive stream errors after the watch has started.
function geoRead(position) {
// position.latitude, position.longitude — WGS84 decimal degrees
// position.accuracy — 95% confidence radius in meters
// position.altitude — meters above ellipsoid (may be null)
// position.altitudeAccuracy — meters (may be null)
// position.heading — degrees from true north (null when stationary)
// position.speed — meters per second (may be null)
// position.timestamp — Epoch ms
}
function onGeoError(error) {
// error.code: 1=PERMISSION_DENIED, 2=POSITION_UNAVAILABLE, 3=TIMEOUT
}Usage:
let startPos = null;
let traveled = 0;
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
// Optional: opt into real GPS (~5-10m outdoors). Comment out for coarse default.
setGeoOptions({ enableHighAccuracy: true, timeout: 30000, maximumAge: 0 });
enableGeoTap('Tap to enable GPS');
}
function draw() {
background(20);
textAlign(CENTER, CENTER);
textSize(20);
const pos = window.lastGeoPosition;
if (window.geoStatus === 'requesting-permission') {
fill(255);
text('Acquiring GPS…', width / 2, height / 2);
text('(cold start can take 5-30s)', width / 2, height / 2 + 30);
} else if (pos) {
fill(120, 220, 160);
text(pos.latitude.toFixed(5) + ', ' + pos.longitude.toFixed(5), width / 2, height / 2 - 30);
text('±' + Math.round(pos.accuracy) + ' m', width / 2, height / 2);
text(traveled.toFixed(1) + ' m traveled', width / 2, height / 2 + 30);
} else if (window.geoError) {
fill(220, 120, 120);
text(window.geoError, width / 2, height / 2, width * 0.8);
} else {
fill(200);
text('GPS not active', width / 2, height / 2);
}
}
function geoRead(position) {
if (!startPos) {
startPos = position; // first fix becomes the origin
} else {
traveled = geoDistance(
startPos.latitude, startPos.longitude,
position.latitude, position.longitude, 'm'
);
}
}Geofencing example — respond when the phone enters a defined area:
const home = [
{ lat: 40.012, lon: -105.260 },
{ lat: 40.012, lon: -105.250 },
{ lat: 40.005, lon: -105.250 },
{ lat: 40.005, lon: -105.260 }
];
function geoRead(position) {
if (geoInPolygon(home, position)) {
background(120, 220, 160); // inside the geofence
} else {
background(60);
}
}Best Practices:
- Always check
window.lastGeoPosition(orwindow.geoStatus) before relying on position data - Use the
geoRead()callback for real-time position processing; readwindow.lastGeoPositionindraw()for display - Start with the coarse default — only opt into
enableHighAccuracy: truewhen you genuinely need GPS-grade accuracy, since it drains more battery and has a slower cold start - Call
stopGeo()when your sketch no longer needs position updates to save battery (it is also called automatically when the sketch is removed in p5.js 2.x) - Serve over HTTPS — geolocation is blocked on insecure origins
- On iOS, GPS does not run when the screen is locked or the tab is backgrounded; expect updates to resume when the page is visible again
- On iOS in-app browsers (Instagram/Facebook), geolocation often fails — instruct users to open the sketch in Safari
Purpose: Send and receive typed values between a p5.js sketch and an Arduino-class BLE peripheral (companion P5PhoneBLE Arduino library). Ideal for sensor dashboards, physical controllers, and bidirectional installations.
Platform Support:
- ✅ Android — Chrome
- ✅ Desktop — Chrome, Edge
- ❌ iOS Safari / Chrome — Web Bluetooth not available; use the free Bluefy browser app on iPhone/iPad
- Requires HTTPS (or
localhost) - Embedded iframes (Canvas LMS, editor previews) need
allow="bluetooth"on the iframe - Peripheral side: pair with the P5PhoneBLE Arduino library — supports UNO R4 WiFi, Nano 33 IoT, Nano 33 BLE, and ESP32 / S3 / C3
Two-step workflow:
- Call
bleSetup()insetup()to declare the service profile (no user gesture needed). - Call
enableBleButton()(or anotherenableBle*helper) so the user can connect from a tap.
Commands:
bleSetup({ serviceUUID, namePrefix, characteristics })— declare typed characteristics (read,write,notify)enableBleTap(options?),enableBleButton(options?),enableBleCanvas(options?),enableBleBanner(options?),enableBleOn(selector)— gesture-gated connect UIbleConnect()— low-level connect (must run inside a user gesture)bleDisconnect()— disconnect from the peripheralbleRead(name)— poll a read-only characteristic (read: trueinbleSetup())bleWrite(name, value, { ack: false })— send a typed value (default: reliable write with response)isBleSupported()— check Web Bluetooth availability
Status variables:
window.bleSupported— browser exposes Web Bluetoothwindow.bleConnected— GATT link is upwindow.bleStatus—'idle' | 'requesting' | 'connecting' | 'connected' | 'disconnected' | 'error' | 'unsupported'window.bleError— latest error messagewindow.bleDeviceName— connected device namewindow.bleValues— latest decoded values keyed by characteristic name (read indraw())
Optional callbacks:
function bleReceive(name, value) { /* fired on each notification */ }
function bleReady(deviceName) { /* fired once connected */ }
function bleClosed() { /* fired on disconnect */ }Usage:
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
bleSetup({
namePrefix: 'p5phone',
characteristics: [
{ name: 'temp', type: 'float', notify: true },
{ name: 'brightness', type: 'uint8', write: true }
]
});
enableBleButton({ label: 'Connect device' });
showDebug();
}
function draw() {
background(20);
if (bleConnected) {
text('temp: ' + (bleValues.temp ?? '—'), 20, 40);
}
}
function mousePressed() {
if (bleConnected) {
bleWrite('brightness', floor(map(mouseX, 0, width, 0, 255)));
}
}Omit characteristic uuid fields to auto-derive them from the service UUID and declaration order — the same contract used by the P5PhoneBLE Arduino library. All numeric types use little-endian byte order.
Best practices:
- Call
bleSetup()before any connect helper - Read
bleValuesindraw(); do not assume a fresh value every frame - Use
showDebug()when testing on phones without devtools - Pair with a matching P5PhoneBLE Arduino sketch (same names, types, and order)
Purpose: Sync plain JSON variables across phones in the same room — multiplayer sketches, shared scoreboards, presence cursors. Backend is a small PartyServer Durable Object you deploy for your project (free Cloudflare Workers tier). Companion: P5PhoneShare.
Platform Support:
- ✅ Any modern mobile/desktop browser with WebSocket (iOS Safari, Android Chrome, desktop)
- Sketch hosting can be GitHub Pages / p5 editor / static hosting; only the sync worker deploys to Cloudflare
Launch a server for your project (once; student or anyone):
cd companion/P5PhoneShare
npm install
npx wrangler login # free Cloudflare account
npx wrangler deploy # prints https://….workers.dev — yours until you delete it- Put that URL in
shareSetup({ host })in your sketch. - Pick a
roomname (any string). Same room = same shared state. - Open the sketch on phones and tap to join.
- Optional:
showDesktopQr()so friends can scan/copy a link withshareHost+roomalready in the URL (they do not need to deploy).
Full student-oriented steps: companion/P5PhoneShare/README.md.
Join links (invite others without them running Wrangler):
https://your-sketch.example/?shareHost=https://YOUR.workers.dev&room=my-game
shareSetup()readsshareHost/room/appfrom the URL (URL wins over config defaults).getShareJoinUrl()returns the current page URL with those params filled in.showDesktopQr()aftershareSetup()encodes that join URL (opt out with{ share: false }).
Sketch workflow:
- Call
shareSetup({ host, room, shared?, me?, app? })insetup()(host/room may come from the URL instead). - Call
enableShareTap()(or anotherenableShare*helper) so the user joins from a tap. On desktop,showDesktopQr()helps others join.
Commands:
shareSetup({ host, room, app, shared, me, autoReconnect, sendInterval, disconnectWhenHidden })— configure PartyServer host and room (URL query params override host/room/app when present)sendInterval(ms, default50): changes are batched and sent at most this often;0sends each change at oncedisconnectWhenHidden(defaulttrueon phones): leave the room while the page is hidden (screen locked, another app) and rejoin when it is visible, keepingme
getShareJoinUrl()— sketch URL withshareHost/room/appfor copy/QR sharingenableShareTap(label | options?),enableShareButton(label | options?),enableShareCanvas(label | options?),enableShareBanner(label | options?),enableShareMinimal(label | options?),enableShareOn(selector)— gesture-gated join UIshareConnect()/shareDisconnect()— low-level connect/disconnectshareSet(path, value)/shareSetMe(path, value)— explicit path writes ('pos.x')shareEmit(name, data?)— room one-shot events (not stored)isShareSupported()— WebSocket availability check
Live objects / status:
shared— room-scoped object (Proxy; assign likeshared.score = 1; arrays sync too:shared.list.push(x))me— this client's per-guest objectguests— array of other clients'meobjectswindow.shareSupported,shareConnected,shareStatus,shareError,shareRoom,shareClientId,shareIsHost
Optional callbacks:
function shareReady() { /* welcome applied */ }
function shareReceive(path, value) { /* remote patch */ }
function shareClosed() {}
function shareHostChanged(isHost) {}
function shareEvent(name, data) { /* shareEmit traffic */ }Minimal sketch:
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
shareSetup({
host: 'https://YOUR.workers.dev', // from: cd companion/P5PhoneShare && npx wrangler deploy
room: 'demo',
shared: { score: 0 },
me: { x: 0.5, y: 0.5 }
});
enableShareTap('Tap to join room');
showDesktopQr(); // optional invite QR / copy-link for friends
}
function draw() {
background(20);
if (!shareConnected) return;
fill(255);
text(shared.score, width / 2, 40);
circle(me.x * width, me.y * height, 40);
for (const g of guests) {
circle(g.x * width, g.y * height, 30);
}
}
function mousePressed() {
if (!shareConnected) return false;
shared.score = (shared.score || 0) + 1;
me.x = mouseX / width;
me.y = mouseY / height;
return false;
}Best practices:
- Deploy your own P5PhoneShare worker for your project (free Cloudflare account)
- To invite others, share a sketch link (or desktop QR) that already contains
shareHost+room - Keep values JSON-serializable (no functions, DOM,
NaN) - Writing
me.x = mouseXevery frame is fine: unchanged values are not sent and changes are batched (sendInterval), which keeps a class inside the free Cloudflare tier - Last write wins. If two phones write the same key at the same moment, every phone ends up with the same value, but one write is lost: two simultaneous
shared.score += 1taps count once. For counts, have each phone keep its own tally inmeand add upguests, or let the host (shareIsHost) own the value - Changes to
sharedmade while disconnected are replaced by the room's state when you rejoin;meis kept - After updating p5-phone, redeploy your worker (
npx wrangler deploy). 1.14.0 uses protocol v2; an older worker answers with an "Unsupported protocol version" error - See companion/P5PhoneShare/PROTOCOL.md for the wire format
Roadmap (not built yet): remember host in localStorage, short room codes only, one-click Cloudflare deploy button, optional Supabase/Vercel adapters.
Purpose: Enable the Web Speech API for voice input and speech-to-text in mobile browsers.
Important: This does NOT create a p5.AudioIn object — it only activates the audio context needed for the Web Speech API. You must create your own p5.SpeechRec object after enabling.
Commands:
enableSpeechTap(message)- Tap anywhere on screen to enable speech recognitionenableSpeechButton(text)- Creates a button with custom text to enable speech
Usage:
let speechRec;
function setup() {
createCanvas(windowWidth, windowHeight);
// Enable speech recognition
enableSpeechTap('Tap to enable speech recognition');
}
// Use userSetupComplete() to know when permissions are ready
function userSetupComplete() {
if (window.speechEnabled) {
speechRec = new p5.SpeechRec('en-US');
speechRec.continuous = true;
speechRec.interimResults = true;
speechRec.onResult = gotSpeech;
speechRec.start();
}
}
function gotSpeech() {
if (speechRec.resultValue) {
debug('You said:', speechRec.resultString);
}
}Purpose: Simplified camera access optimized for ML5.js machine learning models (FaceMesh, HandPose, BodyPose, ObjectDetection, etc.). Handles camera initialization, coordinate mapping, mirroring, and display modes automatically.
Key Features:
- Automatic Coordinate Mapping - ML5 keypoints and bounding boxes automatically mapped to canvas coordinates
- Mirror Support - Handles front camera mirroring for natural interaction
- Display Modes - Multiple video sizing options (fitHeight, cover, contain, fixed)
- ML5 Optimized - Direct integration with ML5 v1.x models
- Auto-initialization - Camera starts automatically when permissions are granted
Commands:
| Function | Purpose | Parameters |
|---|---|---|
createPhoneCamera(active, mirror, mode) |
Create new camera instance | active: 'user' or 'environment' mirror: true/false mode: 'fitHeight', 'cover', 'contain', 'fixed' |
enableCameraTap(message) |
Tap to enable camera | Optional message string |
cam.onReady(callback) |
Execute code when camera ready | Callback function |
cam.mapKeypoint(keypoint) |
Map single ML5 keypoint to screen | ML5 keypoint object |
cam.mapKeypoints(keypoints) |
Map array of ML5 keypoints | Array of ML5 keypoints |
cam.mapBox(box) |
Map single ML5 bounding box to screen | ML5 detection object with x, y, width, height |
cam.mapBoxes(boxes) |
Map array of ML5 bounding boxes | Array of ML5 detection objects |
Properties:
| Property | Description | Type |
|---|---|---|
cam.ready |
Camera initialization status | Boolean |
cam.video |
p5.js video element | p5.Element |
cam.active |
Current camera ('user'/'environment') | String |
cam.mirror |
Mirror state | Boolean |
cam.mode |
Display mode | String |
cam.width |
Video width | Number |
cam.height |
Video height | Number |
Basic Setup:
let cam;
let facemesh;
let faces = [];
function setup() {
createCanvas(windowWidth, windowHeight);
// Create camera: front camera, mirrored, fit to canvas height
cam = createPhoneCamera('user', true, 'fitHeight');
// Enable camera (auto-starts if permission granted)
enableCameraTap();
// Start ML5 when camera is ready
cam.onReady(() => {
let options = {
maxFaces: 1,
refineLandmarks: false,
flipped: false // cam.mapKeypoint() handles mirroring
};
facemesh = ml5.faceMesh(options, modelLoaded);
});
}
function modelLoaded() {
// Start detection - use cam.videoElement for ML5
facemesh.detectStart(cam.videoElement, (results) => {
faces = results;
});
}
function draw() {
background(220);
// Draw camera feed
if (cam.ready) {
image(cam, 0, 0); // PhoneCamera handles positioning automatically
}
// Draw tracked face keypoints
if (faces.length > 0) {
let face = faces[0];
// Map nose tip keypoint (index 1) to screen coordinates
let nose = cam.mapKeypoint(face.keypoints[1]);
// Use coordinates for interaction
fill(255, 0, 0);
circle(nose.x, nose.y, 30);
// Map all keypoints at once
let allPoints = cam.mapKeypoints(face.keypoints);
for (let point of allPoints) {
circle(point.x, point.y, 3);
}
}
}Display Modes:
| Mode | Behavior |
|---|---|
'fitHeight' |
Scale video to canvas height (default, recommended) |
'cover' |
Fill entire canvas (may crop video) |
'contain' |
Fit entire video in canvas (may show letterboxing) |
'fixed' |
Fixed size (set with cam.fixedWidth, cam.fixedHeight) |
Coordinate Mapping:
The mapKeypoint(), mapKeypoints(), mapBox(), and mapBoxes() functions automatically handle:
- Video-to-canvas scaling
- Mirror transformation (for front camera)
- Offset positioning (for different display modes)
- 3D coordinates (preserves z-depth from BlazePose)
// Single keypoint
let nose = cam.mapKeypoint(face.keypoints[1]);
console.log(nose.x, nose.y, nose.z); // Screen coordinates + depth
// Multiple keypoints
let hands = cam.mapKeypoints(hand.keypoints);
hands.forEach(point => {
circle(point.x, point.y, 5);
});
// Object detection bounding box
let mappedObject = cam.mapBox(detection);
rect(mappedObject.x, mappedObject.y, mappedObject.width, mappedObject.height);ML5 Model Examples:
// FaceMesh (468 keypoints)
let options = { maxFaces: 1, refineLandmarks: false, flipped: false };
facemesh = ml5.faceMesh(options, modelLoaded);
// HandPose (21 keypoints per hand)
let options = { maxHands: 2, runtime: 'mediapipe', flipped: false };
handpose = ml5.handPose(options, modelLoaded);
// BodyPose (33 keypoints with 3D)
let options = { modelType: 'MULTIPOSE_LIGHTNING', flipped: false };
bodypose = ml5.bodyPose('BlazePose', options, modelLoaded);
// ObjectDetection (bounding boxes)
objectDetector = await ml5.objectDetection('cocossd');
objectDetector.detectStart(cam.videoElement, gotDetections);Important Notes:
- Always set
flipped: falsein ML5 options when available (PhoneCamera handles mirroring) - Use
cam.videoElement(native HTML video element) when passing to ML5'sdetectStart() - Check
cam.readybefore using video or drawing keypoints - Call
enableCameraTap()to handle camera permissions automatically
Helpers that aren't tied to one piece of hardware: keep the browser from scrolling or zooming, get a sketch from your laptop onto your phone, see errors on the phone itself, and keep the screen on.
Purpose: Prevents unwanted mobile browser gestures that can interfere with your p5.js app.
When to use: Call once in your setup() function after creating the canvas.
Modes:
- fullscreen (default) — page-wide gesture blocking for full-viewport sketches
- embedded — canvas-scoped blocking for sketches inside scrollable multi-page sites
Options:
| Option | Default | Description |
|---|---|---|
mode |
'fullscreen' |
'fullscreen' or 'embedded' |
element |
canvas element | Target element for embedded mode |
warnBeforeLeave |
false |
Show browser "Leave site?" dialog on navigation |
trapHistory |
true in fullscreen |
Block back-swipe via history manipulation |
What it blocks:
- Pinch-to-zoom - Prevents users from accidentally zooming the page
- Pull-to-refresh - Stops the browser refresh gesture when pulling down
- Swipe navigation - Disables back/forward swipe gestures (fullscreen mode)
- Long-press context menus - Prevents copy/paste menus from appearing
- Text selection - Stops accidental text highlighting on touch and hold
- Double-tap zoom - Eliminates double-tap to zoom behavior
// Full-screen mobile sketch (default)
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
}
// Embedded canvas in a scrollable tutorial page
function setup() {
createCanvas(560, 400);
lockGestures({ mode: 'embedded', element: canvas });
}Purpose: Remove gesture blocking listeners and restore saved handlers.
When to use: Call before navigating away in SPAs, or, with p5.js 2.x, rely on automatic cleanup when calling p.remove().
unlockGestures();Purpose: Get a sketch from your laptop onto your phone without typing a URL. showDesktopQr() shows a small floating QR code of the current page on desktop only; on a phone it does nothing, so you can leave it in the sketch.
Commands:
| Function / variable | Purpose |
|---|---|
showDesktopQr(options?) |
Show the QR panel on desktop (no-op on phones) |
setQrUrl(url) |
Point the QR at a different URL |
hideDesktopQr() |
Remove the QR panel |
window.isMobile |
true on phones and tablets (user agent, touch, and coarse-pointer checks, including iPadOS reporting a Mac user agent) |
window.isDesktop |
!isMobile |
Options (all optional): url (defaults to the current page, or the Present link in the Web Editor), position ('top-right' default, 'top-left', 'bottom-right', 'bottom-left'), size (pixels, default 180), label (caption), closable (default true), rememberDismiss (default true: stays hidden for the browser session after closing), share (after shareSetup(), the QR includes the room's join link; false opts out).
In the p5.js Web Editor the sketch runs from a temporary blob: address that no phone can open, so the QR encodes the sketch's Present link instead (https://editor.p5js.org/<user>/full/<id>, the same page as Share › Present). The phone gets the last saved version, so save before scanning. An unsaved sketch has no link yet and shows no QR. On every other host (GitHub Pages, a local server) the QR is the page's own address, as before.
The QR library (qrcodejs) is loaded from a CDN only when the panel is shown on desktop, so phones download nothing extra. If the CDN is blocked, the panel is removed with a console warning.
function setup() {
createCanvas(windowWidth, windowHeight);
enableMicTap('Tap to start');
showDesktopQr({ label: 'Scan to open on your phone' });
}
function draw() {
background(20);
if (window.isDesktop) {
text('Open this sketch on your phone', 20, 30);
}
}Purpose: Essential on-screen debugging system for mobile development where traditional browser dev tools aren't accessible. Provides automatic error catching, timestamped logging, and color-coded messages.
Why use it: Mobile browsers often hide JavaScript errors, making debugging difficult. This system displays all errors, warnings, and custom messages directly on your mobile screen with timestamps and color coding.
Commands:
| Function | Purpose | Example |
|---|---|---|
showDebug() |
Show debug panel and enable error catching | showDebug() |
hideDebug() |
Hide debug panel | hideDebug() |
toggleDebug() |
Toggle panel visibility | toggleDebug() |
debug(...args) |
Log messages (white text) | debug("App started", frameRate()) |
debugError(...args) |
Display errors (red text) | debugError("Connection failed") |
debugWarn(...args) |
Display warnings (yellow text) | debugWarn("Low battery") |
debug.clear() |
Clear all messages | debug.clear() |
Key Features:
- Automatic Error Catching - JavaScript errors automatically displayed with red styling
- Error Location - Shows filename and line number for easy debugging
- Timestamps - All messages include precise timestamps
- Color Coding - Errors (red), warnings (yellow), normal messages (white)
- Mobile Optimized - Touch-friendly interface that works on small screens
- Keyboard Shortcuts - Press 'D' to toggle, 'C' to clear (when debug is enabled)
Critical Setup:
function setup() {
// IMPORTANT: Call showDebug() FIRST to catch setup errors
showDebug();
createCanvas(windowWidth, windowHeight);
// Any errors after this point will be automatically caught and displayed
}Usage Examples:
// Basic logging
debug("Touch at:", mouseX, mouseY);
debug("Sensors enabled:", window.sensorsEnabled);
// Error handling
debugError("Failed to load image");
debugWarn("Frame rate dropping:", frameRate());
// Objects and arrays
debug("Touch points:", touches);
debug({rotation: rotationX, acceleration: accelerationX});Purpose: Stop the phone from dimming and locking while a sketch runs. Only touches reset the phone's auto-lock timer, so a sketch driven by motion, sound, GPS, or BLE goes dark after 30 seconds or so unless it holds a wake lock.
This is the browser's own Screen Wake Lock API, not a p5-phone function. It takes a few lines, so p5-phone does not wrap it.
Commands:
| Command | Purpose |
|---|---|
await navigator.wakeLock.request('screen') |
Keep the screen on while the page is visible. Returns a WakeLockSentinel |
await wakeLock.release() |
Let the screen sleep again. A released sentinel can't be reused; request a new one |
wakeLock.released |
true once the lock is gone, whether the sketch released it or the browser did |
wakeLock.addEventListener('release', fn) |
Run fn when the lock is released |
'wakeLock' in navigator |
false in browsers without the API and on pages not served over HTTPS |
Things to know:
- Ask from a tap, in
mouseReleased(). iOS Safari refuses the first request without a user gesture, and a touch only counts as one when the finger lifts, somousePressed()is too early.userSetupComplete()is too late: it runs after the permission prompts, when iOS may no longer count the tap. Chrome doesn't need a tap, but asking from one works everywhere. - Ask again when the page comes back. The browser releases the lock whenever the page is hidden (another app, another tab, the side button), and it doesn't come back by itself. Request it again on
visibilitychange. On iOS, only the first request needs a tap. - Expect refusals. Power saving or a low battery can refuse or drop the lock, so wrap the request in
try/catch. - HTTPS only (or
localhost). - Not in the p5.js Web Editor. Its preview runs in an iframe without
allow="screen-wake-lock", so the request fails withNotAllowedError. Host the sketch on its own page, such as GitHub Pages. An iframe on your own site needsallow="screen-wake-lock". - It only keeps the screen on. It can't change the phone's auto-lock setting or brightness, turn on a screen that is already off, or keep the sketch running in the background.
Support: Chrome and Edge (Android and desktop), Safari 16.4+ on iOS (Home Screen web apps from iOS 18.4), Firefox 126+.
let wakeLock = null;
let wantAwake = false;
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
// The lock is dropped while the page is hidden. Ask again when it's back.
document.addEventListener('visibilitychange', () => {
if (wantAwake && document.visibilityState === 'visible') {
requestWakeLock();
}
});
}
async function requestWakeLock() {
try {
wakeLock = await navigator.wakeLock.request('screen');
} catch (err) {
debugWarn('Wake lock refused: ' + err.message);
}
}
function mouseReleased() {
const awake = wakeLock !== null && !wakeLock.released;
if ('wakeLock' in navigator && !awake) {
wantAwake = true;
requestWakeLock();
}
return false;
}
function draw() {
const awake = wakeLock !== null && !wakeLock.released;
background(awake ? 'gold' : 20);
}Example: Keep Screen On (tap to toggle the lock, with a timer since the last touch).
Beyond the default Tap and Button styles, p5-phone has four more ways to present permission prompts: Canvas, Banner, and Custom Element (v1.7.0), and Minimal (v1.14.0). Every family has all six: Sensor (or Gyro), Mic, Sound, Speech, Vibration, Torch, Nfc, Geo, Camera, All, Permissions, Ble, and Share. enableBle* and enableShare* take an options object; the others take a message.
The canvas displays a centered message until the user taps. Great for "full-screen tap to start" experiences where you want the canvas to feel like a splash screen.
Commands:
enableSensorCanvas(message),enableMicCanvas(message),enableSoundCanvas(message),enableSpeechCanvas(message)enableVibrationCanvas(message),enableTorchCanvas(message),enableNfcCanvas(message),enableGeoCanvas(message)enableCameraCanvas(message),enableAllCanvas(message),enablePermissionsCanvas(list, message)enableBleCanvas(options),enableShareCanvas(options)
Usage:
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
angleMode(DEGREES);
enableSensorCanvas('Tap canvas to begin');
}
function draw() {
background(220);
if (window.sensorsEnabled) {
circle(width/2 + rotationY * 3, height/2 + rotationX * 3, 50);
}
}A styled banner slides in from the top of the screen with an animated entrance. After the user taps, the banner slides away and permissions are activated. Ideal when you want to keep your canvas visible underneath the prompt.
Commands:
enableSensorBanner(message),enableMicBanner(message),enableSoundBanner(message),enableSpeechBanner(message)enableVibrationBanner(message),enableTorchBanner(message),enableNfcBanner(message),enableGeoBanner(message)enableCameraBanner(message),enableAllBanner(message),enablePermissionsBanner(list, message)enableBleBanner(options),enableShareBanner(options)
Usage:
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
angleMode(DEGREES);
enableSensorBanner('Tap here to enable motion sensors');
}
function draw() {
background(220);
if (window.sensorsEnabled) {
text('Rotation: ' + rotationX.toFixed(1), 20, 60);
}
}Bind the permission activation to any existing HTML element on the page using a CSS selector. This gives you full control over the look and placement of the trigger. The element stays on the page after activation, so hide it yourself once the matching flag is true (see below).
Commands:
enableSensorOn(selector),enableMicOn(selector),enableSoundOn(selector),enableSpeechOn(selector)enableVibrationOn(selector),enableTorchOn(selector),enableNfcOn(selector),enableGeoOn(selector)enableCameraOn(selector),enableAllOn(selector),enablePermissionsOn(selector, list)enableBleOn(selector),enableShareOn(selector)
Usage (HTML):
<button id="start-btn" style="font-size:24px; padding:20px;">
Start Experience
</button>Usage (sketch.js):
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
angleMode(DEGREES);
enableAllOn('#start-btn');
}
function draw() {
background(220);
if (window.sensorsEnabled) {
select('#start-btn').hide(); // the button stays on the page until you hide it
circle(width/2 + rotationY * 3, height/2 + rotationX * 3, 50);
}
}A bare, semi-transparent full-screen overlay with an optional pulsing circle in the center, and no message box. Use it when the frosted Tap box is too much and the sketch should show through. Tapping anywhere activates.
Commands:
enableSensorMinimal(...),enableMicMinimal(...),enableSoundMinimal(...),enableSpeechMinimal(...)enableVibrationMinimal(...),enableTorchMinimal(...),enableNfcMinimal(...),enableGeoMinimal(...)enableCameraMinimal(...),enableAllMinimal(...),enablePermissionsMinimal(list, ...)enableBleMinimal(options),enableShareMinimal(options)
Call forms: enableMicMinimal(), enableMicMinimal('Tap'), enableMicMinimal({ color, opacity, icon, iconColor, iconSize, message }), or enableMicMinimal('Tap', { opacity: 0.6 }).
Options: color (overlay color, default '#000000'), opacity (0 to 1, default 0.5), icon (false hides the pulsing circle), iconColor (default '#ffffff'), iconSize (pixels, default 14), message (optional small text).
function setup() {
createCanvas(windowWidth, windowHeight);
lockGestures();
angleMode(DEGREES);
enableSensorMinimal({ color: '#1a1030', opacity: 0.4, iconColor: '#ffcc00' });
}
function draw() {
background(220);
if (window.sensorsEnabled) {
circle(width/2 + rotationY * 3, height/2 + rotationX * 3, 50);
}
}| Style | Best For | How It Works |
|---|---|---|
| Tap | Quick prototypes | Full-screen transparent overlay |
| Button | Clear UI | Auto-generated styled button |
| Canvas | Splash screens | Message drawn on the p5 canvas |
| Banner | Polished apps | Animated slide-in banner |
| Minimal | Letting the sketch show through | Semi-transparent overlay with a pulsing icon |
| Custom | Custom designs | Bind to your own HTML element |
iOS Safari requires a user gesture (tap, click) before granting access to motion sensors and the microphone. This is a browser security requirement — it cannot be bypassed. Android asks for nothing, but keep the tap there too: in the p5.js Web Editor it is what lets Chrome send motion to the sketch (see the next question).
Since Chrome 153, Chrome pauses the motion sensors in an iframe until the sketch inside it has focus. The Web Editor runs every sketch in an iframe, so tilt and shake did nothing there on Android, while the same sketch on its own page worked. From p5-phone 1.15.3 the motion tap to start (enableGyroTap(), enablePermissionsTap(['gyro', …]) and the others) gives the sketch focus. From 1.15.4 any p5-phone tap does, and so does any later tap or click on the sketch. Load p5-phone@1.15.4 or later, and start motion sketches with a p5-phone tap even though Android needs no permission.
The Web Editor runs every sketch inside an iframe. Almost all of p5-phone works there, once the sketch has had its p5-phone tap to start. That tap gives the sketch focus and the user gesture these features need inside the sketch's own frame.
| Feature | In the Web Editor |
|---|---|
Motion (tilt, shake, deviceShaken()) |
✓ after the tap, with p5-phone 1.15.3 or later (see above) |
| Microphone, sound, speech, camera, torch, GPS, Bluetooth, Share | ✓ after the tap |
| Vibration | ✓ on Android after the tap (iPhones have no vibration anywhere) |
Keyboard (keyPressed()) |
✓ after a tap or click on the sketch. With lockGestures() on, that needs p5-phone 1.15.4 or later |
| NFC | ✗ only works when the sketch is the whole page. window.nfcError says so |
| Screen wake lock | ✗ the editor's iframe does not allow it (see Screen Wake Lock) |
For NFC and the wake lock, open the sketch on its own page, such as GitHub Pages. Other sites that embed sketches in an iframe behave the same way, as long as the iframe's allow attribute lists the feature (for example allow="accelerometer; gyroscope; camera; microphone; geolocation; bluetooth").
- Make sure you're serving over HTTPS — motion sensors and microphone are blocked on insecure origins.
- Check that you've called one of the
enable...functions insetup(). - On iOS, check Settings → Safari → Motion & Orientation Access is enabled.
rotationX/Y/Z follow angleMode(), and p5.js uses radians unless told otherwise, so rotationY * 3 is only a few pixels. Call angleMode(DEGREES) in setup(). See Motion Sensor Activation.
p5.js sets mouseIsPressed to false as soon as any one finger lifts. In a multi-finger sketch, test touches.length > 0 for "a finger is down".
Make sure you're calling enableConsole() in your sketch. The console overlay appears at the bottom of the screen. You can use debug(), debugWarn(), and debugError() to log to it.
Define a userSetupComplete() function — it's called automatically after the user taps and all requested permissions have been granted:
function userSetupComplete() {
debug('All permissions granted!');
debug('Sensors:', window.sensorsEnabled);
debug('Mic:', window.micEnabled);
}You can also check the status variables at any time:
window.sensorsEnabledwindow.micEnabled(request ran) andwindow.micOpen(stream really live)window.speechEnabled
- Ensure your page is served over HTTPS.
- Check that you're using
new PhoneCamera(this)in your sketch. - Grant camera permission in the browser when prompted.
- Some iOS versions require the user to explicitly allow camera access in Settings → Safari → Camera.
Only touches reset the phone's auto-lock timer, so tilting the phone, making sound, or walking with GPS doesn't keep the screen on. Hold a wake lock: see Screen Wake Lock (Keep the Screen On).
The Vibration API is not supported on iOS. Use navigator.vibrate only as an enhancement for Android devices. Check 'vibrate' in navigator before calling it.