ComfyUI-Qwen3TTS-Bridge
Run Qwen3-TTS inside ComfyUI without touching ComfyUI's own Python packages.
Licensed Apache 2.0 at release on GitHub. Since this a resource is for Qwen 3, a different license was added by default here, I'm working to adjust that
Why a bridge?
Current ComfyUI installs ship transformers 5.x. Qwen3-TTS needs transformers 4.57.3:
The official
qwen-ttspackage does not import on transformers 5.Nodes that patch around this can load, but may produce garbled or incoherent audio.
A Python process can only load one version of a package, so both cannot live in ComfyUI's environment. This node keeps Qwen3-TTS in its own virtual environment and runs it as a separate process. The node is the bridge: you work in ComfyUI as usual, and it hands each job to the TTS environment and returns the audio.
Nothing is installed into ComfyUI's environment.
⚠️ Not a copy-and-paste install.
So, the rest of the details are on https://github.com/sd123git/ComfyUI-Qwen3TTS-Bridge/tree/main#comfyui-qwen3tts-bridge rather than copy/pasting here.
This bridge is a baseline, with planned expansion to include the voice design and clone modes.
