PR

Fish Speech S2 Pro Dual AR

ProGGUF by subspecsUpdated 2026-10-11
Download on Hugging Face
Parameters4.95B
Context32K
Layers36
Architecturefish-speech
Smallest quant2.6 GB
Quick answer

Fish Speech S2 Pro Dual AR needs about 4.8 GB of memory at Q4_K_M (a 3.6 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 7 quants, from 2.6 GB (Q2_K) to 9.9 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 3.0 GB.

Fish Speech S2 Pro Dual AR – high‑quality multilingual TTS with fine‑grained prosody control

Fish Speech S2 Pro Dual AR is a 4.95B parameter text‑to‑speech model. It uses a dual‑autoregressive architecture with a 4B slow AR and 400M fast AR. The model supports free‑form inline control tags for prosody and emotion.

Good at

  • Fine‑grained prosody control via inline tags
  • High‑fidelity multilingual speech synthesis
  • Efficient inference with Dual‑AR design

Limits

  • No vision support
  • Requires Vulkan GPU or sufficient CPU RAM
  • License is research‑only; commercial use needs separate license

Best for

  • Voice cloning with short reference audio
  • Custom TTS pipelines needing expressive control
  • Local inference on GPUs with Vulkan support

Who it’s for. Developers building local TTS applications who need expressive, controllable speech synthesis. Researchers interested in multilingual TTS and voice cloning can use the model under the research license.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K2.6 GB 3.8 GB
Fits Get file →
Q3_K3.0 GB 4.3 GB
Fits Get file →
Q4_K_M3.6 GB 4.8 GB
Fits Get file →
Q5_K_M4.0 GB 5.3 GB
Fits Get file →
Q6_K4.5 GB 5.8 GB
Fits Get file →
Q8_05.6 GB 6.9 GB
Fits Get file →
F169.9 GB 11.2 GB
Too big Get file →

For 8 GB VRAM, Q4_K_M (3.57 GB, 4.8 GB memory) is a good balance. If VRAM is limited to 4 GB, Q2_K (2.58 GB, 3.8 GB memory) is the smallest option. For higher quality, Q6_K (4.53 GB, 5.8 GB memory) can be used on 12 GB VRAM systems.

What GPU do you need for Fish Speech S2 Pro Dual AR?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: the best quant that fits is Q2_K (2.6 GB file, about 3.8 GB in use).
  • 6 GB VRAM: the best quant that fits is Q6_K (4.5 GB file, about 5.8 GB in use).
  • 8 GB VRAM: the best quant that fits is Q8_0 (5.6 GB file, about 6.9 GB in use).
  • 12 GB VRAM: the best quant that fits is F16 (9.9 GB file, about 11.2 GB in use).

Which Fish Speech S2 Pro Dual AR quant should you choose?

Q4_K_M is the usual starting point for Fish Speech S2 Pro Dual AR: it balances file size and quality well. Here it is a 3.6 GB file.

The Q8_0 file is 5.6 GB, about 1.6 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is Q2_K at 2.6 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturefish-speech
Parameters4.95B
Layers36
Trained context32,768 tokens
KV cache (fp16)90 KB per token

Source

Quantized bysubspecs
FamilyPro
Licenseother
Downloads (HF)895
Last updated2026-10-11

Frequently asked questions about Fish Speech S2 Pro Dual AR

How much VRAM does Fish Speech S2 Pro Dual AR need?

About 4.8 GB at Q4_K_M with an 8K context window. That is the 3.6 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Fish Speech S2 Pro Dual AR on an 8 GB GPU?

Yes, with Q8_0: the 5.6 GB file needs about 6.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Fish Speech S2 Pro Dual AR GGUF?

It depends on the quant: Q2_K is 2.6 GB, Q3_K is 3.0 GB, Q4_K_M is 3.6 GB, Q5_K_M is 4.0 GB, Q6_K is 4.5 GB, Q8_0 is 5.6 GB and 1 more.

What context length does Fish Speech S2 Pro Dual AR support?

The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.0 GB at 32K.

Where can I download Fish Speech S2 Pro Dual AR GGUF?

The files are on Hugging Face in the subspecs/s2-pro-gguf repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Fish Speech S2 Pro Dual AR use?

Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run Fish Speech S2 Pro on 8 GB VRAM?

Yes, the Q4_K_M quantization requires about 4.8 GB memory at 8k context, which fits in 8 GB VRAM.

Does the model support voice cloning?

Yes, it supports voice cloning with 5–30 second reference audio in WAV or MP3 format.

Is the model free for commercial use?

No, the license allows research and non‑commercial use only; commercial use requires a separate license.

Other GGUFs of s2-pro

Similar models

Browse all →

More from subspecs

All by subspecs →

More Pro models

All Pro →

Other small models

All small models →