sonoranblue/irodori-4bit

Irodori-TTS 600M v3 VoiceDesign — MLX 4-bit

A 4-bit MLX conversion of Aratako/Irodori-TTS-600M-v3-VoiceDesign, a Japanese text-to-speech model by Aratako (Chihiro Arata), for on-device use on Apple silicon.

This is the model that loqi (Sonoran Blue Ltd.) downloads for its Mac version. It works with mlx-audio-swift.

What was changed

• The original weights (model.safetensors) were quantized to 4 bits (group size 64) with mlx_audio.convert. The MLX config.json follows the mlx-community 8-bit conversion (mlx-community/Irodori-TTS-600M-v3-VoiceDesign-8bit).
• The audio codec in dacvae/ is the MLX conversion of Aratako/Semantic-DACVAE-Japanese-32dim (as bundled in the mlx-community conversion), unchanged.
• The tokenizer is not included; use llm-jp/llm-jp-3-150m.

File	Size
model.safetensors	665 MB (4-bit)
model.safetensors.index.json
config.json
dacvae/model.safetensors	430 MB
dacvae/config.json

License

The model weights are released by the original author under the MIT License, and this conversion keeps it.

MIT License

Copyright (c) 2026 Aratako

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

The audio codec (Semantic-DACVAE-Japanese-32dim, MIT, by Aratako) is derived from facebook/dacvae-watermarked (Apache License 2.0, Copyright (c) Meta Platforms, Inc. and affiliates).

Ethical restrictions (from the original model card)

In addition to the license terms, the original author's ethical restrictions apply:

1. No Impersonation: Do not use this model to clone or impersonate the voice of any individual (e.g., voice actors, celebrities, public figures) without their explicit consent.
2. No Misinformation: Do not use this model to generate deepfakes or synthetic speech intended to mislead others or spread misinformation.
3. Voice Generation Disclaimer: When generating speech purely from text or captions without using a reference audio, it is possible that the generated voice may coincidentally resemble that of a real person. This is strictly a probabilistic artifact within the latent space. The model was not trained with the intent of reproducing specific individuals.
4. Liability Disclaimer: The developers assume no liability for any misuse of this model. Users are solely responsible for ensuring their use of the generated content complies with applicable laws and regulations in their jurisdiction.

Credits

• Model: Aratako/Irodori-TTS by Aratako (Chihiro Arata)
• MLX: ml-explore/mlx, mlx-audio by Prince Canuma
• 4-bit conversion: Sonoran Blue Ltd.