Crate svara 

svara — Formant and Vocal Synthesis

svara (Sanskrit: voice/tone/musical note) provides a complete formant-based vocal synthesis pipeline: glottal source generation, vocal tract modeling, phoneme-level synthesis, prosodic control, and sequenced speech rendering.

Architecture

The synthesis pipeline flows:

GlottalSource → VocalTract (FormantFilter + NasalCoupling + LipRadiation) → Output

Higher-level constructs build on this:

• Phonemes define articulatory targets and synthesis strategies per sound class
• Prosody controls f0 contour, stress, and intonation patterns
• Voice profiles parameterize speaker characteristics
• Sequences combine phonemes with coarticulation and crossfading

Quick Start

use svara::prelude::*;

// Create a male voice and synthesize a vowel
let voice = VoiceProfile::new_male();
let samples = synthesize_phoneme(
    &Phoneme::VowelA,
    &voice,
    44100.0,
    0.5,
).expect("synthesis should succeed");

// Or build a phoneme sequence
let mut seq = PhonemeSequence::new();
seq.push(PhonemeEvent::new(Phoneme::VowelA, 0.15, Stress::Primary));
seq.push(PhonemeEvent::new(Phoneme::NasalN, 0.08, Stress::Unstressed));
seq.push(PhonemeEvent::new(Phoneme::VowelI, 0.15, Stress::Secondary));
let audio = seq.render(&voice, 44100.0).expect("render should succeed");

Feature Flags

• naad-backend (default): Use naad crate for oscillators, filters, and noise. Without this, svara uses internal minimal implementations.
• std (default): Enable standard library. Disable for no_std environments (requires alloc).
• logging: Enable tracing-subscriber for structured log output.

Modules

bridge

Dependency-free bridge functions for ecosystem integration.

error

Error types for the svara crate.

formant

Formant filtering and vowel target definitions.

glottal

Glottal source models for vocal synthesis.

lod

Level-of-detail quality control for multi-voice scenarios.

phoneme

Phoneme definitions and synthesis.

pool

Object pooling for transient phoneme synthesis.

prelude

Convenience re-exports for common usage.

prosody

Prosody: pitch contours, intonation patterns, and stress.

render

Async rendering API for non-real-time batch synthesis.

sequence

Phoneme sequencing with coarticulation and crossfading.

spectral

Spectral analysis tools for vocal synthesis.

tract

Vocal tract model connecting glottal source to output.

trajectory

Formant trajectory planning across multi-phoneme windows.

voice

Voice profiles defining speaker characteristics.