dhvani

Core audio engine for Rust.

Buffers, DSP, resampling, mixing, analysis, and capture — in a single crate. The audio equivalent of ranga (image processing) and tarang (media framework).

Name: Dhvani (नाद, Sanskrit) — primordial sound, cosmic vibration. Extracted from shruti (DAW) as a standalone, reusable engine.

What it does

dhvani is the audio processing core — it owns the audio math so nobody else has to. Applications build their audio features on top of dhvani.

Quick start

[dependencies]
dhvani = "0.20"

use dhvani::buffer::{AudioBuffer, mix, resample_linear};
use dhvani::dsp;
use dhvani::analysis;
use dhvani::clock::AudioClock;

// Create buffers
let vocals = AudioBuffer::from_interleaved(samples_a, 2, 44100)?;
let drums = AudioBuffer::from_interleaved(samples_b, 2, 44100)?;

// Mix
let mut mixed = mix(&[&vocals, &drums])?;

// Process
dsp::compress(&mut mixed, 0.5, 4.0);
dsp::normalize(&mut mixed, 0.95);
dsp::noise_gate(&mut mixed, 0.01);

// Analyze
let spectrum = analysis::spectrum_dft(&mixed, 4096);
let loudness = analysis::loudness_lufs(&mixed);
println!("Peak: {:.2}, LUFS: {:.1}", mixed.peak(), loudness);

// Resample for output
let output = resample_linear(&mixed, 48000)?;

// Sync with video via clock
let mut clock = AudioClock::new(48000);
clock.start();
clock.advance(output.frames as u64);
println!("PTS: {} us", clock.pts_us());

Features

# Minimal (no system deps)
dhvani = { version = "0.20", default-features = false }

# With PipeWire capture (Linux)
dhvani = { version = "0.20", features = ["pipewire"] }

Key types

AudioBuffer

Core sample buffer. Holds f32 interleaved samples with channel count and sample rate.

let mut buf = AudioBuffer::silence(2, 44100, 44100); // 1 second stereo
buf.apply_gain(0.5);
buf.clamp();
println!("Peak: {:.3}, RMS: {:.3}, Duration: {:.2}s",
    buf.peak(), buf.rms(), buf.duration_secs());

AudioClock

Sample-accurate transport with tempo awareness and PTS generation for A/V sync.

let mut clock = AudioClock::with_tempo(44100, 120.0); // 120 BPM
clock.start();
clock.advance(44100); // 1 second
println!("Position: {:.2}s, Beat: {:.1}, PTS: {} us",
    clock.position_secs(),
    clock.position_beats().unwrap(),
    clock.pts_us());

DSP

dsp::noise_gate(&mut buf, 0.01);        // silence below threshold
dsp::compress(&mut buf, 0.5, 4.0);      // reduce dynamic range
dsp::hard_limiter(&mut buf, 0.95);       // prevent clipping
dsp::normalize(&mut buf, 1.0);           // peak normalize

let db = dsp::amplitude_to_db(0.5);      // -6.02 dB
let amp = dsp::db_to_amplitude(-6.0);    // ~0.501

Analysis

let spectrum = analysis::spectrum_dft(&buf, 4096);
if let Some(freq) = spectrum.dominant_frequency() {
    println!("Dominant: {:.0} Hz", freq);
}

let lufs = analysis::loudness_lufs(&buf);
let silent = analysis::is_silent(&buf, -60.0);

The Sanskrit Stack

shruti (श्रुति — that which is heard) creates music
  └── with dhvani (नाद — primordial sound) as its audio engine
       └── carried by tarang (तरंग — wave) as its media framework
            └── colored by ranga (रंग — color) for visual processing

Who uses this

Roadmap

Full details: docs/development/roadmap.md

Building from source

git clone https://github.com/MacCracken/dhvani.git
cd dhvani

# Build (no system deps needed)
cargo build

# Build with PipeWire (Linux, requires libpipewire-dev)
sudo apt install libpipewire-0.3-dev
cargo build --features pipewire

# Run tests
cargo test

# Run benchmarks
cargo bench

# Run all CI checks locally
make check

Versioning

Pre-1.0 releases use 0.D.M (day.month) SemVer — e.g. 0.20.3 = March 20th. Post-1.0 follows standard SemVer.

License

AGPL-3.0-only. See LICENSE for details.