---
title: FastDialogueSidon
canonical_url: "https://www.modelscope.cn/models/higurashi-ai/FastDialogueSidon"
md_url: "https://www.modelscope.cn/models/higurashi-ai/FastDialogueSidon.md"
repository: higurashi-ai/FastDialogueSidon
last_updated: 2026-11-16
license: apache-2.0
model_type:
  - fast-dialogue-sidon
parameters: 433.4M
tensor_type:
  - F32
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
  - ja
  - de
  - es
  - fr
  - zh
downloads: 29
stars: 2
tags:
  - speech-separation
  - dialogue
  - speaker-separation
  - w2v-bert
  - triton
  - tensorrt
---

# FastDialogueSidon

> FastDialogueSidon - higurashi-ai 在 ModelScope 开源的模型。Weights for FastDialogueSidon, a fast one-step model for full-duplex two-speaker dialogue separation and joint restoration: up to 120 s of 16 kHz mono dialogue in, two separated 24 kHz speaker tracks out.…

higurashi-ai/FastDialogueSidon 是 ModelScope 魔搭社区上的 433.4M 参数机器学习模型，采用 apache-2.0 许可。

- **Repository**: higurashi-ai/FastDialogueSidon
- **License**: apache-2.0
- **Parameters**: 433.4M
- **Tags**: speech-separation, dialogue, speaker-separation, w2v-bert, triton, tensorrt
- **Downloads**: 29
- **Stars**: 2
- **Last updated**: 2026-11-16

Source: https://www.modelscope.cn/models/higurashi-ai/FastDialogueSidon

---

# FastDialogueSidon

Weights for **FastDialogueSidon**, a fast one-step model for full-duplex two-speaker
dialogue separation and joint restoration: up to 120 s of 16 kHz mono dialogue in,
two separated 24 kHz speaker tracks out. It combines a **one-step DiT head** (a
single DiT step per window instead of the 30 diffusion steps of
[DialogueSidon](https://arxiv.org/abs/2604.09344)) with a **lightweight
decoder** (a frame-rate SAN-WaveNeXt2 (WaveFitNeXt) latent decoder, 30 M parameters, in place
of DialogueSidon's DAC decoder). Every
module — encoder, DiT head and decoder — is trained for FastDialogueSidon.
Inference code with a Triton backend (no engine build) and a TensorRT backend
lives in
[higurashi-ai/FastDialogueSidon](https://github.com/higurashi-ai/FastDialogueSidon).

![FastDialogueSidon architecture](assets/architecture.png)

## Model

| file | params | what |
|---|---|---|
| `encoder/model.safetensors` | 314.6 M | 13-layer w2v-BERT 2.0-architecture encoder with `relative_key` attention, + 2 speaker latent heads |
| `diffusion_head/model.safetensors` | 88.7 M | one-step DiT head (8 layers, hidden 768) |
| `vocoder/model.safetensors` | 30.1 M | lightweight SAN-WaveNeXt2 (WaveFitNeXt) latent decoder (2 iterations, frame-rate ConvNeXt blocks) |
| `stats.safetensors` | — | per-channel latent normalization mean / std |
| `config.json`, `*/config.json` | — | module hyper-parameters and the one-step update constants |

All weights are fp32 safetensors (≈ 1.73 GB in total).

## Usage

```bash
git clone https://github.com/higurashi-ai/FastDialogueSidon && cd FastDialogueSidon
hf download higurashi-ai/FastDialogueSidon --local-dir weights/
cd triton && pip install -e .
python scripts/run.py --bundle ../weights --wav dialogue.wav --output-dir out/   # any length <= 120 s
```

```python
from fds_triton.runtime import TritonPipeline, load_waveform
pipe = TritonPipeline("weights/")
audio = pipe.separate(load_waveform("dialogue.wav"))   # [2, samples] @ 24 kHz
```

The TensorRT backend (`trt/` in the code repository) builds its engines from
the same weights:
`python scripts/build_engines.py --bundle ../weights --engine-dir artifacts/$(hostname)-trt10 --onnx-dir /tmp/fds-onnx`.

## Performance

### Speed

Per-window wall clock on H100 (batch 1) against the original
[DialogueSidon](https://arxiv.org/abs/2604.09344) — official release run as-is
(fp32 PyTorch, DPM-Solver++ 30 steps + DAC) — on the benchmark inputs below:

| | DialogueSidon | FastDialogueSidon (Triton fp16) | speed-up |
|---|---|---|---|
| OpenDialog, 30 s windows | 0.53 s (RTF 0.0178) | 6.2 ms (RTF 2.07e-4) | **86x** |
| CallFriend, 120 s windows | 2.67 s (RTF 0.0222) | 24.8 ms (RTF 2.07e-4) | **107x** |

### Benchmark

Protocol of the DialogueSidon paper (Sec. 4.1) with these deviations:
CallFriend sessions cropped to 120 s (the paper uses 20 s), OpenDialog uses
100 English sessions of exactly 30 s, and intelligibility is scored with
Qwen3-ASR-1.7B instead of Whisper large-v3. Both systems run on the same
inputs. Neither CallFriend nor OpenDialog is used for training; both serve only as test sets.

**OpenDialog** (English, 100 × 30 s)

| | WER (%) ↓ | DNSMOS ↑ | NISQA ↑ |
|---|---|---|---|
| DialogueSidon | 13.21 | **3.327** | **3.757** |
| FastDialogueSidon (Triton fp16) | **10.80** | 3.303 | 3.491 |

**CallFriend** (5 languages × 20 × 120 s)

| p-CER (%) ↓ | de | es | fr | ja | zh | all |
|---|---|---|---|---|---|---|
| DialogueSidon | 11.56 | **15.39** | **21.02** | 16.36 | 20.93 | 16.54 |
| FastDialogueSidon (Triton fp16) | **11.47** | 16.89 | 21.25 | **14.49** | **19.46** | **16.52** |

| | DNSMOS ↑ | NISQA ↑ | speaker sim. ↑ | VAD acc. ↑ |
|---|---|---|---|---|
| DialogueSidon | 3.293 | **3.626** | 0.941 | **0.931** |
| FastDialogueSidon (Triton fp16) | **3.294** | 3.395 | **0.943** | 0.929 |

See the [code repository](https://github.com/higurashi-ai/FastDialogueSidon)
for per-length latencies of both backends (Triton and TensorRT), where the
speed-up comes from, and the numerical-accuracy comparison.

## License

Apache License 2.0.

## Citation

FastDialogueSidon technical report (Rikuto Kotoge): coming soon.

```bibtex
@inproceedings{nakata2026dialoguesidon,
  title={DialogueSidon: Recovering Full-Duplex Dialogue Tracks from In-the-Wild Dialogue Audio},
  author={Nakata, Wataru and Saito, Yuki and Yamauchi, Kazuki and Tsunoo, Emiru and Saruwatari, Hiroshi},
  booktitle={Proceedings of the 27th Annual Meeting of the Special Interest Group on Discourse and Dialogue},
  pages={1--12},
  year={2026}
}
```
