audio_samples_io

Fast and simple audio reading and writing in Rust

Rust crate providing audio file I/O utilities and helpers.

audio_samples_io is the I/O layer of the AudioSamples ecosystem. It builds on top of the representations provided by audio_samples to provide reading, writing, and streaming for WAV and FLAC files.

The AudioSamples Ecosystem

audio_samples
├── core types, signal generation, DSP, analysis
├── audio_samples_io
│   └── file I/O: read and write WAV and FLAC
├── audio_samples_python
│   └── python bindings via PyO3
├── audio_samples_streaming
│   └── streaming pipeline, device I/O, rodio, async
└── audio_samples_ml
    └── STT and TTS integrations

Crate	What it provides	Start here if...
audio_samples	AudioSamples<T>, core types, signal generation, DSP, analysis	You need in-memory audio representations or signal processing primitives
audio_samples_io	Read and write WAV and FLAC files	You need to load or save audio files with minimal setup
audio_samples_python	Python bindings via PyO3 and NumPy interop	You want to use the library from Python or integrate with Python workflows
audio_samples_streaming	Chunk-based streaming, real-time device I/O, async pipelines	You need real-time audio processing or streaming pipelines
audio_samples_ml	Speech-to-text (STT) and text-to-speech (TTS) integrations	You want to integrate transcription or synthesis into your audio pipeline

NOTE: Still under development

Currently supports .wav (including RF64/BW64 64-bit variants for files over 4 GiB), .flac, and .aiff/AIFF-C files. Read paths detect the container from the file's magic bytes, falling back to the extension.

Planned:

• opus / ogg
• mp3

Quick start

Install via:

cargo add audio_samples_io

use std::time::Duration;

use audio_samples_io::error::AudioIOResult;
use audio_samples::{AudioSamples, sample_rate, sine_wave};

pub fn main() -> AudioIOResult<()> {
    // create and write a basic signal and read it back
    let sine_wave: AudioSamples<f32> =
        sine_wave::<f32>(440.0, Duration::from_secs_f64(10.0), sample_rate!(44100), 1.0);
    audio_samples_io::write("./sine_wave.wav", &sine_wave)?;

    let read_sine_wave: AudioSamples<f32> = audio_samples_io::read("./sine_wave.wav")?;
    println!("{:#}", read_sine_wave);
    println!("Duration: {:.1}s", read_sine_wave.duration_seconds());
    Ok(())
}

Benchmarks

Checkout the wav benchmark summary doc for a comparison against Hound and libsndfile. Check out flac benchmark summary

To run the benchmarks (Criterion) reading and writing.

cargo bench --bench wav_rw --features "wav"

cargo bench --bench flac_rw --features "flac"

I have included a small benchmark reported binary for collating benchmarks to a markdown file

cargo run --bin benchmark_reporter --features benchmark_reporting

Output files

• benchmark_report.md — Markdown summary generated by the reporter binary
• benchmark_results.csv — Detailed CSV export of benchmark rows (times in μs)
• benchmark_summary.csv — Executive summary CSV

License

MIT License

---

Contributing

Contributions are welcome. Please submit a pull request and see CONTRIBUTING.md for guidance.

Citing

If you use AudioSamples in research, please cite:

@inproceedings{geraghty2026audio,
  author    = {Geraghty, Jack and Golpayegani, Fatemeh and Hines, Andrew},
  title     = {Audio Made Simple: A Modern Framework for Audio Processing},
  booktitle = {ACM Multimedia Systems Conference 2026 (MMSys '26)},
  year      = {2026},
  month     = apr,
  publisher = {ACM},
  address   = {Hong Kong, Hong Kong},
  doi       = {10.1145/3793853.3799811},
  note      = {Accepted for publication}
}