# VibeVoice **Repository Path**: alan-ding/vibe-voice ## Basic Information - **Project Name**: VibeVoice - **Description**: 本地部署microsoft vibevoice - **Primary Language**: Unknown - **License**: MIT - **Default Branch**: master - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2026-01-04 - **Last Updated**: 2026-03-01 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README ## microsoft vibevoice family https://www.modelscope.cn/collections/VibeVoice-02135dcb17e242 ## readme https://gitee.com/garyvov/vibevoice ## run ``` set HF_ENDPOINT=https://hf-mirror.com python demo/gradio_demo.py --model_path ./models/VibeVoice-1.5b ``` ## Usage 1: Launch Gradio demo ```bash export HF_ENDPOINT=https://hf-mirror.com apt update && apt install ffmpeg -y # for demo # For 1.5B model python demo/gradio_demo.py --model_path ./models/VibeVoice-1.5b # For Large model python demo/gradio_demo.py --model_path ./models/VibeVoice-Large ``` ## Usage 2: Inference from files directly ```bash # We provide some LLM generated example scripts under demo/text_examples/ for demo # 1 speaker python demo/inference_from_file.py --model_path ./models/VibeVoice-1.5b --txt_path demo/text_examples/1p_abs.txt --speaker_names Alice # or more speakers python demo/inference_from_file.py --model_path ./models/VibeVoice-Large --txt_path demo/text_examples/2p_music.txt --speaker_names Alice Frank ``` ## enable cuda 1. uv remove torch 2. delete torch from .venv 3. use global cuda torch - include-system-site-packages = true 4. reopen project #### VibeVoice-1.5b ![alt text](image-2.png) ![alt text](image-3.png) #### VibeVoice-Large ![alt text](image-4.png) #### VibeVoice-7B ## test ![alt text](image.png) ![alt text](image-1.png)