README.md

Marvis TTS

A fast conversational text-to-speech (TTS) model with built-in voices for English, French, and German.

Hugging Face Model Repo

Supported Voices

• conversational_a (English)
• conversational_b (English)
• conversational_fr (French)
• conversational_de (German)

CLI Example

mlx-audio-swift-tts --model Marvis-AI/marvis-tts-250m-v0.2-MLX-8bit --voice conversational_a --text "Hello world."

Swift Example

import MLXAudioTTS

let model = try await MarvisTTSModel.fromPretrained("Marvis-AI/marvis-tts-250m-v0.2-MLX-8bit")
let audio = try await model.generate(
    text: "Hello world.",
    voice: "conversational_a",
    refAudio: nil,
    refText: nil,
    language: nil,
    generationParameters: GenerateParameters()
)