Crate audiobox 

Audio decoding, editing, analysis and encoding in safe Rust, with zero dependencies, and the audio side of a game: a real-time mixer, an 8-bit synthesizer and parametric sound effects.

The crate uses only the standard library: no C libraries, FFmpeg, build scripts or unsafe. It compiles to WebAssembly, and its synthesizer, effects and mixer give the same samples on every platform.

use audiobox::{AudioFormat, NormalizeOptions, decode, encode};

fn prepare(bytes: &[u8]) -> audiobox::Result<Vec<u8>> {
    let clip = decode(bytes)?
        .cut(1.5, 30.0)?
        .to_mono()
        .resample(16_000)?
        .normalize(NormalizeOptions::default())?;
    encode(&clip, AudioFormat::Flac)
}

What is included

Area	Items	Feature
Decoding and encoding	decode, encode, read_audio_file, write_audio_file for WAV, AIFF, CAF, AU, FLAC and MP3	always
Editing and DSP	Audio: cut, concat, mix, fades, filters, resampling, limiter, tempo and pitch	always
Analysis	Audio::loudness, peaks, RMS, analyze::spectrum, waveforms	always
Streaming	stream::Decoder for WAV, FLAC and MP3 a chunk at a time with seeking; raw PCM and resampling across chunks	always
Disk streaming	stream::FileStream: a native reader thread and bounded PCM/encoded buffers	file-stream
Synthesizer	synth: text scores rendered in an 8-bit style	synth
Sound effects	sfx: parameters, text form and seeded presets	sfx
Game mixer	mixer: voices, buses, fades, loops and an allocation-free render	mixer
Sound assets	assets: background loading and a budgeted decoded-audio cache	assets
Adaptive music	music: beat/bar transitions, crossfades, cues and intensity layers	music
Sound events	events: named text events with seeded choices, variation and limits	events

Enable the optional modules you use:

[dependencies]
audiobox = { version = "0.5", features = ["mixer", "synth", "sfx"] }

Determinism

These give the same samples on every platform, including WebAssembly: the synthesizer, the sound effects and the mixer, and the offline transforms that render audio — Audio::pan, Audio::gain and Audio::gain_db, Audio::fade, Audio::crossfade_to, Audio::filter, Audio::limit, Audio::limit_true_peak, Audio::resample, Audio::tempo, Audio::pitch, Audio::trim_silence, Audio::normalize, and analyze::from_db. They compute with the crate’s own sine and 2^x series instead of the platform’s sin, cos, exp and powf, whose last bits can differ between systems; CI pins their output with checksums on Linux, macOS, Windows and WebAssembly. Analysis that only reports numbers, such as loudness, true peak and spectra, use the platform’s math. Normalization in every unit uses a separate deterministic measurement path and reports its values through Audio::normalize_with_report. Audio::measure_deterministic exposes that path for non-mutating final validation. Both LUFS paths require >= 4000 Hz.

Integer PCM export supports optional seeded Dither::Tpdf noise through EncodeOptions::dither. Its quantization is deterministic, including FLAC and stream::PcmEncoder::with_dither across chunk boundaries. Dither is disabled by default; float, G.711 and MP3 output reject enabled dither.

Errors are Error values with a stable ErrorCode. Decoding is bounded by Limits, checked before allocating. The website has a browser sandbox and a sound-effect and chiptune playground that run this crate through WebAssembly.

Re-exports

pub use mixer::events;

pub use mixer::music;

pub use metadata::CoverArt;

pub use metadata::Tags;

Modules

analyze

Waveforms, spectral analysis and K-weighted integrated loudness.

assets

Background loading and a cache of decoded sounds. Requires assets.

metadata

Publishing tags for MP3 (ID3v2.4), FLAC and WAV INFO.

mixer

A real-time mixer for games. Requires the mixer feature.

mp3

MPEG-1/2/2.5 Layer III decoding, probing and MPEG-1 CBR encoding.

pcm

Raw PCM conversion, including G.711 A-law and µ-law.

sfx

A parametric sound-effect generator in the style of the synth module. Requires the sfx feature, which enables synth.

stream

Incremental decoding and encoding.

synth

An 8-bit style synthesizer that renders text scores. Requires the synth feature.

Structs

Audio

Planar floating-point audio. All channels have the same length. Samples may exceed ±1 to retain headroom; encoders clamp at quantization.

DecodeInfo

DecodeOptions

EncodeOptions

Error

An error with a stable category and human-readable context.

FilterOptions

LimiterOptions

Limits

Allocation and duration limits checked before allocating decoded PCM.

NormalizeOptions

NormalizeReport

ReadAudioOptions

Limits for the whole-file convenience reader. Encoded bytes and decoded PCM have separate budgets; codec-specific limits still apply during decoding.

SilenceOptions

SilentRange

TruePeakLimiterOptions

Opt-in offline true-peak limiter. The existing crate::LimiterOptions and Audio::limit keep their sample-peak behavior.

Enums

AudioFormat

Dither

Optional noise at final linear integer PCM quantization.

Endian

ErrorCode

Stable error categories. Match on the code rather than the message.

FadeCurve

FilterType

NormalizeUnit

SampleFormat

TimePosition

Seconds, an exact frame position, or a parsed timestamp. Negative positions count from the end.

Functions

decode

decode_with_options

encode

encode_with_options

read_audio_file

Decode a regular file using a 512 MiB encoded-byte cap and default decoded crate::Limits. See read_audio_file_with_options for path and error policy.

read_audio_file_with_options

Read a bounded regular file, then decode the same buffer without copying it.

sniff

Detect a supported format from its bytes, independent of the extension.

write_audio_file

Type Aliases

AudioMetadata

Result