ComfyUI Voxtral TTS

ComfyUI custom nodes for Mistral's Voxtral TTS

Nodes

Voxtral TTS Fixed Voice

Generate speech using one of 30 preset voices across 4 speakers and multiple emotions.

Input	Type	Description
text	STRING	Text to synthesize
voice	COMBO	Preset voice selection

Available voices:

• EN - Jane — Sarcasm, Confused, Shameful, Sad, Neutral, Jealousy, Frustrated, Curious, Confident
• EN - Paul — Sad, Neutral, Happy, Frustrated, Excited, Confident, Cheerful, Angry
• EN - Oliver — Neutral, Sad, Excited, Curious, Confident, Cheerful, Angry
• FR - Marie — Sad, Neutral, Happy, Excited, Curious, Angry

Output: AUDIO

---

Voxtral TTS Voice Clone

Clone any voice by providing a 5–30 second reference audio clip.

Input	Type	Description
text	STRING	Text to synthesize
reference_audio	AUDIO	Reference clip (5–30s, WAV/MP3)

Output: AUDIO

Installation

1. Clone or copy this repository: git clone https://github.com/marduk191/comfyui-voxtral.git
2. pip install gradio_client and soundfile from requirements.txt
3. Find the nodes under the audio/voxtral category

If you use the git installer through comfyui manager then you can just install from git and restart.

Requirements

• Internet connection
• gradio_client
• soundfile