KokoroTTS-iOS

Swift Package for text-to-speech on iOS using the Kokoro TTS model.

Note: Currently, the model is adapted and tested only for English (en_us, en_gb) and French (fr) languages. All other languages require preprocessor improvements. Contributors are welcome!

Installation

Add the package to your project via Swift Package Manager:

dependencies: [
    .package(url: "https://github.com/Otosaku/KokoroTTS-iOS", from: "1.0.0")
]

Or in Xcode: File → Add Package Dependencies... and enter the repository URL.

Required Resources

Before using the library, you need to download the following resources:

Resource	Description	URL
TTS Models	Core ML models for speech synthesis	Download
G2P Vocab	Grapheme-to-phoneme vocabulary files	Download
POS Model	Part-of-speech tagging model	Download
Espeak Data	Espeak-ng phoneme data	Download

Usage

Initialize the Pipeline

import iOS_TTS
import CoreML

// Configure ML compute units (optional)
let configuration = MLModelConfiguration()
configuration.computeUnits = .all // or .cpuAndNeuralEngine, .cpuOnly

// Initialize the pipeline
let pipeline = try TTSPipeline(
    modelPath: modelsDirectory,           // URL to extracted TTS models
    vocabURL: vocabDirectory,             // URL to extracted G2P vocab
    postaggerModelURL: posModelsDirectory, // URL to extracted POS model
    language: .englishUS,                  // Language selection
    espeakDataPath: espeakDataDirectory.path, // Path to espeak-ng data
    configuration: configuration
)

Generate Speech

// Basic generation
let audioSamples = try await pipeline.generate(text: "Hello, world!")

// With options
let options = GenerationOptions(
    style: .afHeart,  // Voice style
    speed: 1.0        // Speech speed
)
let audioSamples = try await pipeline.generate(text: "Hello, world!", options: options)

Available Languages

Language	Code	Status
English (US)	.englishUS	Supported
English (GB)	.englishGB	Supported
French	.french	Supported
Spanish	.spanish	Needs work
Italian	.italian	Needs work
Portuguese	.portuguese	Needs work
Hindi	.hindi	Needs work
Japanese	.japanese	Needs work
Chinese	.chinese	Needs work

Available Voices

English (US): afHeart, afAlloy, afAoede, afBella, afJessica, afKore, afNicole, afNova, afRiver, afSarah, afSky, amAdam, amEcho, amEric, amFenrir, amLiam, amMichael, amOnyx, amPuck, amSanta

English (GB): bfAlice, bfEmma, bfIsabella, bfLily, bmDaniel, bmFable, bmGeorge, bmLewis

French: ffSiwis

Example Project

The repository includes an Example app that demonstrates:

• Downloading and extracting required resources
• Initializing the TTS pipeline
• Generating and playing speech
• Voice and language selection

To run the example:

1. Open Example/Example.xcodeproj in Xcode
2. Build and run on an iOS device

Requirements

• iOS 16.0+
• Swift 6.0+
• Xcode 16.0+

License

MIT License