aufklarer/Qwen3-TTS-CoreML

Qwen3-TTS — CoreML

CoreML conversion of Qwen/Qwen3-TTS-0.6B for Apple Neural Engine acceleration. Includes the codec LM, Mimi decoder, and code embedder as separate CoreML models.

Models

Model	Description
CodeDecoder.mlmodelc	Mimi audio codec decoder
CodeEmbedder.mlmodelc	Token embedding layer
Additional .mlmodelc	Transformer layers for the codec LM

Usage

Used by speech-swift Qwen3TTSCoreML module:

let model = try await Qwen3TTSCoreMLModel.fromPretrained()
let audio = model.synthesize(text: "Hello world", language: "english")

---

• Guide: soniqo.audio/guides/speak
• Docs: soniqo.audio
• GitHub: soniqo/speech-swift