audio-buffer

Audio data container with planar float32 layout.

Web Audio API AudioBuffer ponyfill for node, bun and other envs.

Comes with optional utils.

import AudioBuffer from 'audio-buffer'
import { from, trim, normalize } from 'audio-buffer/util'

let buf = from([0, 0, 0.5, 0.8, 0.3, 0, 0])
buf = normalize(trim(buf))

API

Constructor

new AudioBuffer({ length: 1024, sampleRate: 44100, numberOfChannels: 2 })
new AudioBuffer(2, 1024, 44100) // positional form

Properties

All read-only.

• length — samples per channel
• sampleRate — Hz
• duration — seconds
• numberOfChannels

Iterable over channels: for (let ch of buf), let [L, R] = buf.

Methods

buf.getChannelData(0)                           // Float32Array for channel 0
buf.copyFromChannel(dest, 0, 100)               // copy channel 0 from sample 100 into dest
buf.copyToChannel(src, 1)                       // write src into channel 1

Utils

import { from, slice, concat, set } from 'audio-buffer/util'

Same-size ops mutate and return the buffer. Size-changing ops return a new buffer.

isAudioBuffer(buf) → boolean

Check if object is an AudioBuffer instance or duck-typed compatible.

isAudioBuffer(buf)                               // true

from(source, fill?, options?) → AudioBuffer

Create buffer from anything — number, Float32Array, Array, AudioBuffer, ArrayBuffer. Optional second arg fills samples — a number for constant value, or fn(sample, index, channel, channelData) to map (like Array.from).

from([0.1, -0.3, 0.5])                       // array of samples → mono
from([left, right], { sampleRate: 48000 })   // Float32Array[] → stereo
from(existingBuffer)                         // clone
from(1024, 0.5)                              // 1024 samples buffer filled with 0.5
from(1024, (s, i) => Math.sin(i * 0.1))      // generate sine wave
from(buf, v => v * 0.5)                      // clone with half volume
from(buf, 0)                                 // same shape, zeroed (like)
from({ length: 1024, numberOfChannels: 2 })  // from options

slice(buffer, start?, end?) → newBuffer

Extract sample range into new buffer.

slice(buf, 100, 200)                           // samples 100–199
slice(buf, -50)                                // last 50 samples

concat(...buffers) → newBuffer

Join buffers (same sampleRate and channels).

concat(a, b)                                   // a + b end-to-end
concat(a, b, c, d)                             // join many

set(buffer, other, offset?) → buffer

Overwrite samples from another buffer.

set(buf, patch, 1000)                          // write patch at sample 1000

fill(buffer, value, start?, end?) → buffer

Fill with constant or per-sample function.

fill(buf, 0)                                   // silence
fill(buf, (s, i, ch) => Math.sin(i * 0.1))     // sine wave
fill(buf, () => Math.random() * 2 - 1)         // white noise
fill(buf, v => -v)                             // phase-invert

normalize(buffer, start?, end?) → buffer

Peak-normalize to 1.0, preserving inter-channel balance.

normalize(buf)                                 // quiet recording → full scale

trim(buffer, threshold?) → newBuffer

Remove silence from both ends.

trim(buf)                                      // remove exact zeros
trim(buf, 0.01)                                // remove near-silence

reverse(buffer, start?, end?) → buffer

Reverse samples.

reverse(buf)                                   // full reverse
reverse(buf, 0, 100)                           // reverse first 100 samples

mix(a, b, ratio?, offset?) → a

Blend b into a. Ratio 0 = keep a, 1 = replace with b.

mix(track, reverb, 0.3)                        // 70% dry, 30% wet
mix(a, b, (sa, sb) => Math.max(sa, sb))       // custom blend function
mix(a, b, 0.5, 1000)                           // mix starting at sample 1000

remix(buffer, channels, interpretation?) → newBuffer

Upmix/downmix channels per Web Audio spec speaker rules.

remix(stereo, 1)                               // stereo → mono
remix(mono, 2)                                 // mono → stereo
remix(buf, 6)                                  // → 5.1 surround
remix(buf, 4, 'discrete')                      // copy channels, silence rest

pad(buffer, length, value?, side?) → newBuffer

Pad to target length.

pad(buf, 44100)                                // zero-pad to 1 second
pad(buf, 44100, 0, 'start')                    // pad at start

repeat(buffer, times) → newBuffer

Repeat N times.

repeat(buf, 4)                                 // loop 4x

rotate(buffer, offset) → buffer

Circular shift. Positive = right.

rotate(buf, 100)                               // shift right 100 samples
rotate(buf, -50)                               // shift left 50

removeDC(buffer, start?, end?) → buffer

Remove DC offset (subtract mean per channel).

removeDC(buf)                                  // center waveform at zero

isEqual(a, b) → boolean

Deep equality — same shape, same samples.

isEqual(buf, clone)                            // true if identical

Play

import play from 'audio-buffer/play'

let ctrl = await play(buf, { volume: 0.8, loop: true })
ctrl.pause()
ctrl.play()
ctrl.stop()
ctrl.currentTime  // seconds
ctrl.playing      // boolean

Options: { volume, loop, start, end, autoplay, onended }.

Uses Web Audio API in browsers. In Node.js, install audio-speaker: npm i audio-speaker.

stop() resets to start. play() restarts from beginning.

Replaces

• audio-buffer-utils
• audio-buffer-from
• audio-buffer-remix
• audio-play