meomeo 0.1.1

Text to speech for Dart, powered by espeak-ng and ONNX Runtime.

meomeo

Text to speech for Dart. Text in, audio out.

Uses espeak-ng for phonemization and ONNX Runtime for neural inference. Compatible with KittenTTS models.

Setup

1. Download a KittenTTS model

From HuggingFace, download both files into your project:

Nano (15M params, fast):

curl -L -o model.onnx https://huggingface.co/KittenML/kitten-tts-nano-0.8/resolve/main/kitten_tts_nano_v0_8.onnx
curl -L -o voices.npz https://huggingface.co/KittenML/kitten-tts-nano-0.8/resolve/main/voices.npz

Mini (80M params, better quality):

curl -L -o model.onnx https://huggingface.co/KittenML/kitten-tts-mini-0.8/resolve/main/kitten_tts_mini_v0_8.onnx
curl -L -o voices.npz https://huggingface.co/KittenML/kitten-tts-mini-0.8/resolve/main/voices.npz

2. Compile espeak phoneme data

The espeak package needs phoneme data at runtime. Add espeak as a dev dependency and run the compiler:

dart pub add espeak --dev
dart run espeak:compile_data --all --exclude=fo --output ./espeak-data

Note: dart pub global activate espeak_cli does not work yet because global activate doesn't build native assets. Use dart run from a project that depends on espeak instead.

3. Add meomeo to your project

dependencies:
  meomeo: ^0.1.0

Usage

import 'package:meomeo/meomeo.dart';

final meo = Meo.init(
  'model.onnx',
  voicesPath: 'voices.npz',
  espeakDataPath: './espeak-data',
);

// Text to audio
final pcm = meo.speak('Hello world');

// Text to WAV file
meo.save('Hello world', 'output.wav');

// Switch voice
meo.speaker = 'Bruno';

// Cleanup
meo.dispose();

Voices

Bella, Jasper, Luna, Bruno, Rosie, Hugo, Kiki, Leo

Requirements

• Rust toolchain (rustup) — dort compiles ONNX Runtime via Native Assets
• C compiler (Xcode/gcc) — espeak compiles espeak-ng via Native Assets