Fish Audio Drama 3 ComfyUI Workflows (text to speech + voice clone)
===================================================================

Article: https://freakyjolly.com/fish-audio-drama-3-local-comfyui/

What's inside:
  - drama3_text_to_speech.json: ComfyUI workflow (UI format). Drag it onto the ComfyUI canvas or use Workflow > Open.
  - drama3_voice_clone.json: ComfyUI workflow (UI format). Drag it onto the ComfyUI canvas or use Workflow > Open.

Required models (download each file and put it in the folder shown):
  - drama3_codec.safetensors
      folder: ComfyUI/models/vae/
      download: https://huggingface.co/PedroMarinhoDev/Drama-3-ComfyUI/resolve/main/drama3_codec.safetensors
  - drama3_int8experts_bf16.safetensors
      folder: ComfyUI/models/diffusion_models/
      download: https://huggingface.co/PedroMarinhoDev/Drama-3-ComfyUI/resolve/main/drama3_int8experts_bf16.safetensors
  - qwen3_asr_1.7b_bf16.safetensors
      folder: ComfyUI/models/asr/
      download: https://huggingface.co/PedroMarinhoDev/Qwen3-ASR-ComfyUI/resolve/main/qwen3_asr_1.7b_bf16.safetensors

Custom nodes:
  - ComfyUI-Drama3: https://github.com/PedroMarinhoDev/ComfyUI-Drama3  (ComfyUI Manager > Install Missing Custom Nodes, or git clone into ComfyUI/custom_nodes/)

How to use:
  1. Install ComfyUI-Drama3 and restart ComfyUI (ComfyUI 0.40 or newer).
  2. Download the model files into the folders listed. The workflows load drama3_int8experts_bf16 (the file the article recommends under 24 GB VRAM); pick drama3_fp8experts_bf16, w6a8 or w4a8 in Load Drama 3 instead if that suits your RAM.
  3. Drag drama3_text_to_speech.json (plain TTS) or drama3_voice_clone.json (clone a voice from a clip + transcript) onto the canvas.
  4. Type your script, connect a voice if you want a specific one, and queue.

Notes:
  - Only clone voices you own or have permission to use.

Credits and licence:
  - Official example workflows from PedroMarinhoDev/ComfyUI-Drama3 (GPL-3.0): https://github.com/PedroMarinhoDev/ComfyUI-Drama3. Only the model file selection was changed (fp8 -> int8 experts).
  - Drama 3 weights: Fish Audio licence, see https://huggingface.co/fishaudio/drama-3.

Model weights are not included. Each model keeps its own licence; check the model page before commercial use.
Workflow shared by FreakyJolly.com.
