AEmotionStudio/fish-speech-s2-pro

Fish Speech S2 Pro — Mirror

Mirror of the Fish Speech S2 Pro model by Fish Audio.

Original model: fishaudio/fish-speech-1.5

Available Files

File	Size	Description
model.safetensors	9.12 GB	Main language model weights
codec.pth	1.87 GB	Audio codec (encoder/decoder)
config.json	1.86 KB	Model configuration
tokenizer.json	12.2 MB	Tokenizer data
tokenizer_config.json	861 KB	Tokenizer configuration
special_tokens_map.json	102 KB	Special tokens mapping
chat_template.jinja	4.12 KB	Chat template

Model Details

Fish Speech is a leading open-source text-to-speech (TTS) model that supports high-quality voice cloning and multilingual speech synthesis. The S2 Pro variant offers improved quality and zero-shot voice cloning capabilities.

• Architecture: Qwen3-based language model + audio codec
• Task: Text-to-speech, voice cloning
• Languages: English, Chinese, Japanese, and more
• Code: github.com/fishaudio/fish-speech

Usage with ComfyUI-FFMPEGA

This model is automatically downloaded and used by the ComfyUI-FFMPEGA extension for TTS and voice cloning features.

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: business@fish.audio)