dracon-tts-runtime

Text-to-speech with Kitten (espeak-ng) and Kokoro (ONNX) backends.

Usage

use dracon_tts_runtime::{KittenTTS, KokoroTts};

// Lightweight TTS
let kitten = KittenTTS::new().await?;
kitten.speak("Hello world")?;

// Neural TTS (higher quality)
let kokoro = KokoroTts::new(model_path, voices_dir).await?;
kokoro.speak("Hello world")?;

Feature Flags

• kitten — espeak-ng-based TTS (default, no GPU required)
• kokoro — ONNX neural TTS (requires ort ONNX runtime)

API Notes

All public methods that can fail return anyhow::Result<T>. Constructors (new, new_with_voice) are async and must be awaited with ? or .await?.

Key Types

• [KittenTTS] — espeak-ng backend
• [KokoroTts] — ONNX Kokoro backend
• [TtsEngine] — enum for runtime engine selection
• resolve_voice(name) — resolves voice aliases ("male", "female", etc.)
• [TtsResult<T>][crate::contracts::TtsResult] — anyhow::Result<T> type alias

License

MIT OR Apache-2.0