Confucius4-TTS-mlx

Full-precision (fp32) MLX build of netease-youdao/Confucius4-TTS (multilingual + cross-lingual zero-shot voice cloning, 14 languages: zh, en, ja, ko, de, fr, es, id, it, th, pt, ru, ms, vi) for Apple Silicon.

All weights kept in fp32 (T2S, w2v-bert encoder, S2A flow, BigVGAN vocoder). ~5.4 GB total. For a smaller/faster 8-bit build see mlx-community/Confucius4-TTS-mlx-int8.

• T2S: ~2.5 GB
• w2v-bert: ~1.5 GB
• S2A + BigVGAN: ~0.8 GB
• Speed (Apple M5): RTF ~2.4

Usage

Needs the confucius4 model in mlx-audio (PR #799):

from mlx_audio.tts.utils import load
model = load("mlx-community/Confucius4-TTS-mlx")
for r in model.generate("Xin chào", ref_audio="voice.wav", lang="vi"):
    ...  # r.audio at 22050 Hz

Attribution & license

• Model & architecture: netease-youdao/Confucius4-TTS (Apache-2.0)
• Vocoder: NVIDIA BigVGAN v2; speaker encoder: 3D-Speaker CAMPPlus (funasr)
• MLX port by Hert4, released under Apache-2.0.