Qwen3 TTS 12Hz 0.6B Base — MLX 8-bit

MLX 8-bit quantized conversion of Qwen/Qwen3-TTS-12Hz-0.6B-Base for Apple Silicon inference.

Usage

Used by speech-swift Qwen3TTS module:

let model = try await Qwen3TTSModel.fromPretrained(
    modelId: "aufklarer/Qwen3-TTS-12Hz-0.6B-Base-MLX-8bit"
)
let audio = try model.synthesize("Hello, world!")

audio speak "Hello, world!" --model base-8bit -o output.wav

Model Details

• Architecture: Qwen3-TTS (Talker transformer + Code Predictor + Mimi speech tokenizer decoder)
• Parameters: 0.6B
• Quantization: 8-bit (MLX, talker + code predictor)
• Size: ~1.3 GB
• Sample rate: 24 kHz
• Codec rate: 12.5 Hz

Variants

---

---

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