Hujiazeng / Vach

Real time streaming talking head

Vach: Real-Time stream talking head

流式数字人，实现音视频同步对话,基本可以达到商用效果

来自群友@不蠢不蠢 部署成功的视频展示

Features

• 文本交互
• 语音交互
• SyncTalk项目支持
• 声音克隆
• 直播间业务
• 展厅显示屏互动

Installation

Tested on Ubuntu 18.04, Pytorch 1.12.1 and CUDA 11.3.

git clonehttps://github.com/Hujiazeng/Vach.git
cd Vach

Install dependency

conda create -n Vach python==3.10
conda activate Vach
pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu113
pip install -r requirements.txt
pip install --no-index --no-cache-dir pytorch3d -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu113_pyt1121/download.html
# Note the following modules. If installation is unsuccessful, you can navigate to the path and use pip install . or python setup.py install to compile and install.
# NeRF/freqencoder
# NeRF/gridencoder
# NeRF/raymarching
# NeRF/shencoder

数字人模型文件

我们提供预训练模型下载并测试

可以替换成自己训练的模型(https://github.com/Fictionarry/ER-NeRF)

.
├── data
│   ├── obama(user-defined)
│       ├── transforms_train.json
│       ├── au.csv			
│       ├── ngp_kf.pth
│       ├── template.npy(首次运行自动生成)
│       ├── torso_imgs(仅全身推理时使用)
│       ├── fullbody_imgs(仅全身推理时使用)

Quick Start

python app.py

开启麦克风监听功能

python app.py --mike

如果访问不了huggingface，在运行前

export HF_ENDPOINT=https://hf-mirror.com

用浏览器打开http://127.0.0.1:8010/webrtc.html, 建立连接后, 在文本框提交任何文字。

如果项目对你有帮助，帮忙点个star。也欢迎感兴趣的朋友一起来完善该项目。

微信：hairong0907 加我进交流群

Acknowledgement

This code is developed heavily relying on aiortc, and also ER-NeRF and SyncTalk.

Thanks for these great projects.