tonejs

SKILL.md

Tone.js Skill

Build interactive music applications in the browser using the Web Audio API through Tone.js's high-level abstractions.

When to Use This Skill

Use Tone.js when:

  • Creating synthesizers, samplers, or musical instruments
  • Building step sequencers, drum machines, or DAWs
  • Adding sound effects or music to games
  • Implementing audio visualizations synchronized to sound
  • Processing audio in real-time with effects
  • Scheduling musical events with precise timing
  • Working with musical concepts (notes, tempo, measures)

Core Concepts

1. Context and Initialization

The AudioContext must be started from a user interaction (browser requirement):

import * as Tone from "tone";

// ALWAYS call Tone.start() from user interaction
document.querySelector("button").addEventListener("click", async () => {
	await Tone.start();
	console.log("Audio context ready");
	// Now safe to play audio
});

2. Audio Graph and Routing

All audio nodes connect in a graph leading to Tone.Destination (the speakers):

// Basic connection
const synth = new Tone.Synth().toDestination();

// Chain through effects
const synth = new Tone.Synth();
const filter = new Tone.Filter(400, "lowpass");
const delay = new Tone.FeedbackDelay(0.125, 0.5);
synth.chain(filter, delay, Tone.Destination);

// Parallel routing (split signal)
const reverb = new Tone.Reverb().toDestination();
const delay = new Tone.Delay(0.2).toDestination();
synth.connect(reverb);
synth.connect(delay);

3. Time and Scheduling

Tone.js abstracts time in musical notation:

  • "4n" = quarter note
  • "8n" = eighth note
  • "2m" = two measures
  • "8t" = eighth note triplet
  • Numbers = seconds

CRITICAL: Always use the time parameter passed to callbacks:

// CORRECT - sample-accurate timing
const loop = new Tone.Loop((time) => {
	synth.triggerAttackRelease("C4", "8n", time);
}, "4n");

// WRONG - JavaScript timing is imprecise
const loop = new Tone.Loop(() => {
	synth.triggerAttackRelease("C4", "8n"); // Will drift
}, "4n");

4. Transport System

The global timekeeper for synchronized events:

// Schedule events on the Transport
const loop = new Tone.Loop((time) => {
	synth.triggerAttackRelease("C4", "8n", time);
}, "4n").start(0);

// Control the Transport
Tone.Transport.start();
Tone.Transport.stop();
Tone.Transport.pause();
Tone.Transport.bpm.value = 120; // Set tempo

Step-by-Step Instructions

Task 1: Create a Basic Synthesizer

  1. Import Tone.js
  2. Create a synth and connect to output
  3. Wait for user interaction to start audio
  4. Play notes using triggerAttackRelease
import * as Tone from "tone";

const synth = new Tone.Synth().toDestination();

button.addEventListener("click", async () => {
	await Tone.start();
	// Play C4 for an eighth note
	synth.triggerAttackRelease("C4", "8n");
});

Task 2: Create a Polyphonic Instrument

  1. Use PolySynth to wrap a monophonic synth
  2. Pass multiple notes to play chords
  3. Release specific notes when needed
const polySynth = new Tone.PolySynth(Tone.Synth).toDestination();

// Play a chord
polySynth.triggerAttack(["C4", "E4", "G4"]);

// Release specific notes
polySynth.triggerRelease(["E4"], "+1");

Task 3: Load and Play Audio Files

  1. Create a Player or Sampler
  2. Wait for Tone.loaded() promise
  3. Start playback
const player = new Tone.Player("https://example.com/audio.mp3").toDestination();

await Tone.loaded();
player.start();

// For multi-sample instruments
const sampler = new Tone.Sampler({
	urls: {
		C4: "C4.mp3",
		"D#4": "Ds4.mp3",
		"F#4": "Fs4.mp3",
	},
	baseUrl: "https://example.com/samples/",
}).toDestination();

await Tone.loaded();
sampler.triggerAttackRelease(["C4", "E4"], 1);

Task 4: Create a Looping Pattern

  1. Use Tone.Loop or Tone.Sequence for patterns
  2. Pass the time parameter to instrument triggers
  3. Start the loop and the Transport
const synth = new Tone.Synth().toDestination();

const loop = new Tone.Loop((time) => {
	synth.triggerAttackRelease("C4", "8n", time);
}, "4n").start(0);

await Tone.start();
Tone.Transport.start();

Task 5: Add Effects Processing

  1. Create effect instances
  2. Connect in desired order (serial or parallel)
  3. Adjust wet/dry mix if needed
const synth = new Tone.Synth();
const distortion = new Tone.Distortion(0.4);
const reverb = new Tone.Reverb({
	decay: 2.5,
	wet: 0.5, // 50% effect, 50% dry
});

synth.chain(distortion, reverb, Tone.Destination);

Task 6: Automate Parameters

  1. Access parameter via property (e.g., frequency, volume)
  2. Use methods like rampTo, linearRampTo, exponentialRampTo
  3. Schedule changes with time parameter
const osc = new Tone.Oscillator(440, "sine").toDestination();

osc.start();

// Ramp frequency to 880 Hz over 2 seconds
osc.frequency.rampTo(880, 2);

// Set value at specific time
osc.frequency.setValueAtTime(440, "+4");

// Exponential ramp (better for frequency)
osc.frequency.exponentialRampTo(220, 1, "+4");

Task 7: Synchronize Visuals with Audio

  1. Use Tone.Draw.schedule() for visual updates
  2. Schedule in the same callback as audio events
  3. Visual updates happen just before audio plays
const loop = new Tone.Loop((time) => {
	synth.triggerAttackRelease("C4", "8n", time);

	// Schedule visual update
	Tone.Draw.schedule(() => {
		element.classList.add("active");
	}, time);
}, "4n");

Common Patterns

Pattern: Step Sequencer

const synth = new Tone.Synth().toDestination();
const notes = ["C4", "D4", "E4", "G4"];

const seq = new Tone.Sequence(
	(time, note) => {
		synth.triggerAttackRelease(note, "8n", time);
	},
	notes,
	"8n"
).start(0);

Tone.Transport.start();

Pattern: Probabilistic Playback

const loop = new Tone.Loop((time) => {
	if (Math.random() > 0.5) {
		synth.triggerAttackRelease("C4", "8n", time);
	}
}, "8n");

Pattern: Dynamic Effect Parameters

const filter = new Tone.Filter(1000, "lowpass").toDestination();
const lfo = new Tone.LFO(4, 200, 2000); // 4Hz, 200-2000Hz range

lfo.connect(filter.frequency);
lfo.start();

Sound Design Principles

Core Insights

Auditory processing is 10x faster than visual (~25ms vs ~250ms). Sound provides immediate feedback that makes interactions feel responsive. A button that clicks feels faster than one that doesn't, even with identical visual feedback.

Sound communicates emotion instantly. A single tone conveys success, error, or tension better than visual choreography. When audio and visuals tell the same story together, the experience is stronger than either alone.

Less is more. Most interactions should be silent. Reserve sound for moments that matter: confirmations for major actions, errors that can't be overlooked, state transitions, and notifications. Always pair sound with visuals for accessibility - sound enhances, never replaces. Study games for reference - they've perfected informative, emotional, non-intrusive audio feedback.

Design Philosophy

Good sound design transforms user experience across all platforms - web apps, mobile apps, desktop applications, and games. These principles apply universally whether creating notification sounds, UI feedback, or musical interactions.

Sound uses a universal language understood by everyone. When designing audio:

Ask foundational questions:

  • What is the essence of what this app/feature is about?
  • What emotion do you want to evoke?
  • How does it match the app's visual aesthetics?
  • How would users understand this interaction without looking at the screen?

Consider context:

  • Where will users hear this? (Pocket, desk, busy street, quiet room)
  • What will they be doing? (Working, commuting, gaming)
  • How often will they hear it? (Once per day vs hundreds of times)

Notification Sound Design

Effective notification sounds have these characteristics:

1. Distinguishable

  • Create a unique sonic signature that identifies the app
  • Use characteristic timbres or melodic patterns
  • Layer simple elements to build recognition
  • Don't mimic system defaults or other common sounds

2. Conveys meaning

  • The sound should connect to the message (not literal, but suggestive)
  • Liquid qualities for water/weather, metallic for alerts, warm tones for success
  • Use timbre and envelope to suggest the content
  • Abstract representation, not sound effects

3. Friendly and appropriate

  • Match urgency to message importance
  • Gentle sounds: Soft attacks (50ms+), smooth timbres (sine, triangle)
  • Urgent sounds: Fast attacks (<5ms), brighter timbres (square, FM synthesis)
  • Volume and brightness indicate priority

4. Simple and clean

  • Avoid complex layering or dense harmonic content
  • One or two-note patterns work better than melodies
  • Pleasant timbres remain tolerable when heard repeatedly
  • Clarity over cleverness

5. Unobtrusive and repeatable

  • Duration: 0.3-0.8 seconds maximum for notifications
  • Use softer timbres (sine, triangle) for frequent sounds
  • Avoid harsh, complex, or extremely bright timbres
  • Should remain pleasant when heard 50+ times per day

6. Cuts through noise, not abrasive

  • Mid-range frequencies (300-3000Hz) are most effective
  • Avoid extreme highs (>8kHz) and lows (<80Hz)
  • Design for noisy environments without being harsh
  • Triangle and sine waves are gentler than square waves

UI Sound Design

For buttons, interactions, and transitions:

1. Use sparingly

  • Add sound only to significant state changes or confirmations
  • Silence is often the best choice
  • Don't sonify every hover, mouseover, or minor interaction
  • Reserve audio feedback for meaningful moments

2. Volume relative to purpose

  • UI sounds: Much quieter than notifications (-20 to -30dB)
  • Users have device in hand, subtlety works
  • Notifications: Louder to cut through ambient noise (-10 to -15dB)

3. Synchronization matters

  • Audio and visual updates must occur simultaneously (< 10ms tolerance)
  • Misalignment feels laggy and unprofessional
  • Use precise scheduling, not approximate timing

4. Match interaction character

  • Quick taps: Short duration (<50ms), higher frequencies
  • Heavy presses: Longer duration (100-200ms), lower frequencies
  • Drag operations: Continuous feedback or silence
  • Successful actions: Rising pitch or bright timbre
  • Failed actions: Falling pitch or dull timbre

5. Convey depth and movement

  • Forward movement: Rising pitch over time
  • Backward movement: Falling pitch over time
  • Opening: Expanding envelope (fade in)
  • Closing: Contracting envelope (fade out)
  • Envelope shape suggests physical metaphor

Creative Process

1. Start with a sound palette

  • Explore different synthesis types (sine, triangle, square, FM, AM, noise)
  • Record or synthesize sounds related to the concept
  • Physical objects, everyday sounds, and traditional instruments all work
  • Build a library of 5-10 candidate sounds

2. Match sound to purpose

  • Each UI element gets ONE sound that fits its specific purpose
  • Button click: Quick and responsive (high pitch, fast decay)
  • Toggle switch: Two distinct sounds for on/off states
  • Notification: Longer and more distinctive
  • No need to provide multiple options - choose the best fit

3. Use any sound source

  • Traditional instruments: Kalimba, bells, piano, marimba
  • Physical materials: Metal, glass, wood, plastic
  • Synthesized: Pure tones, FM synthesis, filtered noise
  • Unconventional: Toasters, doors, mechanical sounds
  • The source matters less than the final character

4. Layer for richness

  • Combine 2-3 simple elements rather than one complex sound
  • Base tone + harmonic layer + texture
  • Each layer serves a purpose (body, brightness, character)
  • Keep total duration brief even with layers

Technical Considerations

1. Clean audio

  • Fade in/out to prevent clicks (10-50ms)
  • Remove DC offset
  • Normalize levels consistently
  • Avoid truncation or abrupt endings

2. Frequency filtering

  • High-pass filter at 80Hz (removes sub-bass)
  • Small speakers can't reproduce low frequencies
  • Low-pass filter at 8kHz if needed (removes harsh highs)
  • Focus energy in 300-3000Hz for maximum effectiveness

3. Cross-platform design

  • Design for the worst speaker system (phone, laptop)
  • Headphones reveal more detail, but optimize for speakers
  • Avoid sub-bass (<80Hz) entirely
  • Mid-range frequencies work everywhere

4. Duration guidelines

  • Notifications: 0.3-0.8 seconds
  • Button clicks: 0.05-0.1 seconds
  • Toggles: 0.05-0.15 seconds
  • Transitions: 0.1-0.3 seconds
  • Keep all UI sounds brief to respect user attention

5. User control

  • Always provide settings to disable sounds
  • Volume control separate from system volume
  • Per-sound-type toggles (notifications vs UI feedback)
  • Respect system-wide mute switches

6. Synchronization precision

  • Schedule audio at exact same moment as visuals
  • 10-20ms latency is perceptible and feels wrong
  • Use precise timing APIs, not setTimeout/setInterval
  • Audio-visual sync is critical for perceived responsiveness

Design Checklist

Implementation considerations when designing sounds:

  • Distinguishable: Unique waveform/envelope that doesn't mimic system defaults
  • Repeatable: Simple, pleasant timbres (sine, triangle) rather than complex or harsh
  • Cross-platform: Frequencies between 300-3000Hz, high-pass filtered at 80Hz
  • Audible but not harsh: Mid-range frequencies, avoid extreme highs (>8kHz) and lows (<80Hz)
  • Short duration: 0.3-0.8 seconds for notifications, <0.1 seconds for UI feedback
  • Synchronized: Audio triggers scheduled at same time as visual updates (< 10ms tolerance)
  • User control: Settings option to disable sounds and adjust volume
  • Appropriate character: Envelope matches interaction (quick tap = short decay, heavy press = longer decay)
  • Clean audio: Fade in/out to prevent clicks, filtered to remove unused frequencies
  • Meaningful: Timbre/pitch suggests the message (liquid for rain, metallic for alerts, warm for success)

Important Edge Cases and Gotchas

1. Browser Autoplay Policy

MUST call Tone.start() from user interaction. Without this, no audio will play.

// WRONG - will fail silently
Tone.Transport.start();

// CORRECT
button.addEventListener("click", async () => {
	await Tone.start();
	Tone.Transport.start();
});

2. Memory Management

Always dispose of nodes when done:

const synth = new Tone.Synth().toDestination();

// When finished
synth.dispose();

// For arrays of instruments
players.forEach((player) => player.dispose());

3. Timing Precision

JavaScript callbacks are NOT precise. Always use the time parameter:

// WRONG - will drift out of sync
setInterval(() => {
	synth.triggerAttackRelease("C4", "8n");
}, 250);

// CORRECT - sample-accurate
new Tone.Loop((time) => {
	synth.triggerAttackRelease("C4", "8n", time);
}, "4n").start(0);

4. Loading Samples

Wait for samples to load before playing:

const sampler = new Tone.Sampler({
	urls: { C4: "piano.mp3" },
	baseUrl: "/audio/",
}).toDestination();

// WRONG - may not be loaded yet
sampler.triggerAttack("C4");

// CORRECT
await Tone.loaded();
sampler.triggerAttack("C4");

5. Monophonic vs Polyphonic

Basic synths are monophonic (one note at a time):

// Only plays one note
const mono = new Tone.Synth().toDestination();
mono.triggerAttack(["C4", "E4", "G4"]); // Only C4 plays

// Plays all notes
const poly = new Tone.PolySynth(Tone.Synth).toDestination();
poly.triggerAttack(["C4", "E4", "G4"]); // All play

6. Note Format

Notes can be specified multiple ways:

synth.triggerAttackRelease("C4", "8n"); // Pitch-octave notation
synth.triggerAttackRelease(440, "8n"); // Frequency in Hz
synth.triggerAttackRelease("A4", "8n"); // A4 = 440Hz

7. Transport Time vs Audio Context Time

Two timing systems exist:

Tone.now(); // AudioContext time (always running)
Tone.Transport.seconds; // Transport time (starts at 0)

// Schedule on AudioContext
synth.triggerAttackRelease("C4", "8n", Tone.now() + 1);

// Schedule on Transport
Tone.Transport.schedule((time) => {
	synth.triggerAttackRelease("C4", "8n", time);
}, "1m");

Architecture Overview

ToneAudioNode (base class)
├── Source (audio generators)
│   ├── Oscillator, Player, Noise
│   └── Instrument
│       ├── Synth, FMSynth, AMSynth
│       ├── Sampler
│       └── PolySynth
├── Effect (audio processors)
│   ├── Filter, Delay, Reverb
│   ├── Distortion, Chorus, Phaser
│   └── PitchShift, FrequencyShifter
├── Component (building blocks)
│   ├── Envelope, Filter, LFO
│   └── Channel, Volume, Panner
└── Signal (parameter automation)
    ├── Signal, Add, Multiply
    └── Scale, WaveShaper

Quick Reference

Instrument Types

  • Tone.Synth - Basic single-oscillator synth
  • Tone.FMSynth - Frequency modulation synthesis
  • Tone.AMSynth - Amplitude modulation synthesis
  • Tone.MonoSynth - Monophonic with filter and envelope
  • Tone.DuoSynth - Two-voice synth
  • Tone.MembraneSynth - Percussive synth
  • Tone.MetalSynth - Metallic sounds
  • Tone.NoiseSynth - Noise-based synthesis
  • Tone.PluckSynth - Plucked string model
  • Tone.PolySynth - Polyphonic wrapper
  • Tone.Sampler - Multi-sample instrument

Common Effects

  • Tone.Filter - Lowpass, highpass, bandpass, etc.
  • Tone.Reverb - Convolution reverb
  • Tone.Delay / Tone.FeedbackDelay - Echo effects
  • Tone.Distortion - Waveshaping distortion
  • Tone.Chorus - Chorus effect
  • Tone.Phaser - Phaser effect
  • Tone.PitchShift - Real-time pitch shifting
  • Tone.Compressor - Dynamic range compression
  • Tone.Limiter - Brick wall limiter

Time Notation

  • "4n" - Quarter note
  • "8n" - Eighth note
  • "16n" - Sixteenth note
  • "2m" - Two measures
  • "8t" - Eighth note triplet
  • "1:0:0" - Bars:Beats:Sixteenths
  • 0.5 - Seconds (number)

Resources

Weekly Installs
79
GitHub Stars
1
First Seen
Jan 23, 2026
Installed on
opencode75
gemini-cli74
codex74
cursor73
github-copilot73
amp70