---
title: Irodori-TTS-v4.1-Small-8bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/Irodori-TTS-v4.1-Small-8bit"
md_url: "https://www.modelscope.cn/models/mlx-community/Irodori-TTS-v4.1-Small-8bit.md"
repository: mlx-community/Irodori-TTS-v4.1-Small-8bit
last_updated: 2026-08-15
pipeline_tag: text-to-speech
tasks:
  - text-to-speech
model_type:
  - irodori_tts
base_model:
  - Aratako/Irodori-TTS-v4.1-Small
base_model_relation: finetune
parameters: 321.6M
tensor_type:
  - F32
  - F16
  - U32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 7
stars: 0
tags:
  - mlx
  - japanese
---

# Irodori-TTS-v4.1-Small-8bit

> Irodori-TTS-v4.1-Small-8bit - mlx-community 在 ModelScope 开源的模型。mlx-community/Irodori-TTS-v4.1-Small-8bit

mlx-community/Irodori-TTS-v4.1-Small-8bit 是 ModelScope 魔搭社区上的 321.6M 参数text-to-speech模型，基于 Aratako/Irodori-TTS-v4.1-Small 构建。

- **Repository**: mlx-community/Irodori-TTS-v4.1-Small-8bit
- **Tasks**: text-to-speech
- **Parameters**: 321.6M
- **Base model**: Aratako/Irodori-TTS-v4.1-Small
- **Tags**: mlx, japanese
- **Downloads**: 7
- **Stars**: 0
- **Last updated**: 2026-08-15

Source: https://www.modelscope.cn/models/mlx-community/Irodori-TTS-v4.1-Small-8bit

---

# mlx-community/Irodori-TTS-v4.1-Small-8bit

This model was converted to MLX format from [Aratako/Irodori-TTS-v4.1-Small](https://huggingface.co/Aratako/Irodori-TTS-v4.1-Small) using [mlx-audio version **0.4.7**](https://github.com/Blaizzy/mlx-audio).

## Use with mlx-audio

```bash
pip install -U mlx-audio
```

### Command line

```bash
mlx_audio.tts.generate --model mlx-community/Irodori-TTS-v4.1-Small-8bit --text "こんにちは、Irodori TTSのMLX版です。" --ref_audio reference.wav
```

### Python

```python
from mlx_audio.tts.generate import generate_audio

generate_audio(
    text="こんにちは、Irodori TTSのMLX版です。",
    model="mlx-community/Irodori-TTS-v4.1-Small-8bit",
    ref_audio="reference.wav",
    file_prefix="output",
)
```

## Notes

- Unified v4.1 model: voice cloning, VoiceDesign (caption) and automatic duration
  prediction in one checkpoint.
- v4.1 differs from v4 only in the duration predictor, which upstream retrained
  separately with every other parameter frozen. The remaining 683 of 714 tensors
  are bit-identical to v4.
- The ModernBERT-ja-310m text encoder weights and its tokenizer are bundled, and
  the Semantic-DACVAE-Japanese-32dim codec ships in `dacvae/`, so inference needs
  no additional downloads.
- Reference audio: up to 120s. Passing a list of clips encodes each separately
  and concatenates them, which matches training better than one long recording.
- Short caption-only prompts are shorter than in v4, but still over-predicted.
  Measured with the same reference clip and caption:

  | Text | Tokens | v4 caption only | v4.1 caption only |
  |---|---|---|---|
  | こんにちは。 | 3 | 3.64s | 2.88s |
  | 今日はいい天気ですね。 | 5 | 5.60s | 4.72s |
  | MLXへの移植が完了しました。 | 7 | 4.08s | 3.84s |

  Under about seven tokens the model can still fill the surplus by reading the
  sentence a second time. This matches the reference PyTorch implementation,
  which predicts 117.41 frames against MLX's 117.56 for the second row, and
  produces the same repeat. Pass reference audio, or set `duration_scale` /
  `seconds`.
- Quantized to 8 bits with `mlx.nn.quantize` (group size 64). Note that this
  covers more layers than the upstream torchao checkpoints, which keep AdaLN,
  the projectors and the duration predictor in BF16. Duration predictions stay
  within one frame (40 ms) of the fp16 model.