Downloads · 30 days
47
6% of all-time downloads
fish-speech-s2-pro-fp8
AEmotionStudio/fish-speech-s2-pro-fp8
fish-speech-s2-pro-fp8 is a text-to-speech model from AEmotionStudio. Use it when you need text read aloud. The card lists the license as other.
FP8 weight-only quantization of Fish Audio S2 Pro.
All-time downloads
748
Public
Parameters
4.6B
7 GB on disk
Likes
1
Public
Repo makeup
Click a slice to open those files.
.safetensors5.1 GB · 73%
Parameter types
How the weights are stored.
F8_E4M34B · 89%
Base models
Model card
From the Hugging Face model README
Fish Audio S2 Pro — FP8 (AEmotionStudio Mirror)
FP8 weight-only quantization of Fish Audio S2 Pro.
Details
| Property | Value |
|---|---|
| Source model | fishaudio/s2-pro |
| Quantization | Per-row symmetric FP8 (float8_e4m3fn) |
| Linear layers quantized | 201 |
| FP8 params | 4.048B |
| BF16 params | 0.514B |
| Model size | 4.73 GB |
| VRAM requirement | ~12 GB |
How it works
All nn.Linear weight matrices are quantized to float8_e4m3fn with per-row
float32 scale factors. Non-linear weights (embeddings, layer norms, codec) remain
in bfloat16. No external quantization library is needed — dequantization is pure PyTorch:
W_bf16 = W_fp8.to(torch.bfloat16) * scale
Usage with ComfyUI-FFMPEGA
This model is automatically downloaded and used by the ComfyUI-FFMPEGA extension for TTS and voice cloning features when FP8 precision is selected.
License
Fish Audio Research License — see LICENSE file.
- ✅ Free for research and non-commercial use
- ❌ Commercial use requires a separate license from Fish Audio (contact: [email protected])
Built with Fish Audio.
- Task
- Text-to-Speech
- Type
- fish_qwen3_omni
- License
- other
- Languages
- en, zh
- Created
- Mar 21, 2026
- Updated
- Mar 21, 2026