dhvani 1.0.0

Core audio engine — buffers, DSP, resampling, mixing, analysis, and capture for Rust
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154

//! Grapheme-to-phoneme (G2P) conversion — text to phoneme sequences for vocal synthesis.
//!
//! Re-exports from [`shabda`](https://crates.io/crates/shabda), the AGNOS G2P crate,
//! and [`shabdakosh`](https://crates.io/crates/shabdakosh) pronunciation dictionaries.
//! Converts text strings into [`PhonemeEvent`] sequences ready for voice synthesis.
//!
//! # Feature: `g2p`
//!
//! Enable with:
//! ```toml
//! dhvani = { version = "0.22", features = ["g2p"] }
//! ```
//!
//! # Data Flow
//!
//! ```text
//! text → G2PEngine → Vec<PhonemeEvent> → PhonemeSequence → voice_synth::render_sequence → AudioBuffer
//! ```

use crate::buffer::AudioBuffer;

// ── Engine ─────────────────────────────────────────────────────────

/// G2P conversion engine with language-specific rules and dictionary lookup.
pub use shabda::engine::{G2PEngine, Language};

// ── Rules ──────────────────────────────────────────────────────────

/// Grapheme-to-phoneme rule set.
pub use shabda::rules;

// ── Text normalization ─────────────────────────────────────────────

/// Text normalization (numbers, abbreviations, punctuation).
pub use shabda::normalize;

// ── Syllabification ────────────────────────────────────────────────

/// Syllable boundary detection.
pub use shabda::syllable;

// ── Prosody ────────────────────────────────────────────────────────

/// Prosody assignment from text structure (stress, phrasing).
pub use shabda::prosody;

// ── Dictionary ─────────────────────────────────────────────────────

/// ARPABET phoneme mappings.
pub use shabda::arpabet;

/// Pronunciation dictionary (CMUdict-based).
pub use shabda::dictionary;

// ── Phoneme events (re-exported from svara via shabda) ─────────────

/// Timed phoneme event (phoneme + duration + stress).
pub use svara::sequence::PhonemeEvent;

/// Ordered phoneme sequence with coarticulation.
pub use svara::sequence::PhonemeSequence;

// ── Error ──────────────────────────────────────────────────────────

/// Shabda error type.
pub use shabda::error::ShabdaError;

// ── Bridge: text → dhvani AudioBuffer ──────────────────────────────

/// Convert text to phonemes using the G2P engine.
///
/// # Errors
///
/// Returns `crate::NadaError::Dsp` if G2P conversion fails.
///
/// # Example
///
/// ```rust,no_run
/// use dhvani::g2p::*;
///
/// let engine = G2PEngine::new(Language::English);
/// let phonemes = text_to_phonemes(&engine, "hello world").unwrap();
/// ```
pub fn text_to_phonemes(engine: &G2PEngine, text: &str) -> crate::Result<Vec<PhonemeEvent>> {
    engine
        .convert(text)
        .map_err(|e| crate::NadaError::Dsp(format!("G2P conversion failed: {e}")))
}

/// Speak text directly to a dhvani [`AudioBuffer`].
///
/// Combines G2P conversion and voice synthesis in one call.
///
/// # Errors
///
/// Returns `crate::NadaError::Dsp` if G2P or synthesis fails.
///
/// # Example
///
/// ```rust,no_run
/// use dhvani::g2p::*;
/// use dhvani::voice_synth::VoiceProfile;
///
/// let engine = G2PEngine::new(Language::English);
/// let voice = VoiceProfile::new_female();
/// let buf = speak(&engine, "hello", &voice, 44100).unwrap();
/// ```
pub fn speak(
    engine: &G2PEngine,
    text: &str,
    voice: &svara::voice::VoiceProfile,
    sample_rate: u32,
) -> crate::Result<AudioBuffer> {
    let samples = engine
        .speak(text, voice, sample_rate as f32)
        .map_err(|e| crate::NadaError::Dsp(format!("speak failed: {e}")))?;
    AudioBuffer::from_interleaved(samples, 1, sample_rate)
        .map_err(|e| crate::NadaError::Dsp(format!("buffer from speech output: {e}")))
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn english_g2p() {
        let engine = G2PEngine::new(Language::English);
        let phonemes = text_to_phonemes(&engine, "hello").unwrap();
        assert!(!phonemes.is_empty());
    }

    #[test]
    fn speak_hello() {
        let engine = G2PEngine::new(Language::English);
        let voice = svara::voice::VoiceProfile::new_female();
        let buf = speak(&engine, "hello", &voice, 44100).unwrap();
        assert!(buf.frames() > 0);
        assert!(buf.rms() > 0.0);
        assert!(buf.samples().iter().all(|s| s.is_finite()));
    }

    #[test]
    fn phonemes_to_sequence() {
        let engine = G2PEngine::new(Language::English);
        let phonemes = text_to_phonemes(&engine, "test").unwrap();
        let mut seq = PhonemeSequence::new();
        for p in phonemes {
            seq.push(p);
        }
        let voice = svara::voice::VoiceProfile::new_male();
        let buf = crate::voice_synth::render_sequence(&seq, &voice, 44100).unwrap();
        assert!(buf.frames() > 0);
    }
}