🚀 Turn text into expressive, natural speech — multi-speaker dialogues, emotion tags, and voice cloning from short samples.
▶️ Run Directly in Cloud:
https://www.runcomfy.com/comfyui-workflows/fish-audio-s2-tts-in-comfyui-emotion-multi-speaker-cloning?utm_source=civitai
💡 Overview
Fish Audio S2 TTS is a ComfyUI workflow for advanced text-to-speech: expressive voices, emotion and style tagging, multi-speaker scenes, and precise voice cloning from reference clips.
Perfect for narration, character dialogue, and emotionally rich audio without a recording studio.
✨ Key Features
Multi-speaker: Split scripts across different voices in one graph.
Emotion & style tags: Whispers, laughs, and more for lifelike delivery.
Voice cloning: Match a speaker from a short sample clip.
Fast inference: Iterate quickly on tone and pacing.
🚀 Getting Started
Enter your script and assign speakers / voices as needed.
Add emotion or style hints where you want extra expressiveness.
Generate and export audio for your project.
Click the "Run Directly" link above to bypass local setup and test this workflow immediately in your browser.
Description
Initial release.
FAQ
Comments (2)
If I understand the notes in the workflow correctly, I need at least 16 GB of VRAM. Unfortunately, that's too much—too bad.
I've been looking for something like this! Unfortunately, I keep getting errors. "WhisperConfig object has no attribute 'max_length'" or batch file too big stuff. Neither of these issues seem to have variables that I can find in the workflow to tweak. Anyone have any ideas?