搭建musetalk数字人的步骤

生成数字人的视频效果

搭建步骤

下载git代码

git clone https://github.com/TMElyralab/MuseTalk.git

创建conda环境

(建议使用 python 版本 >=3.10 和 cuda 版本 =11.7。)

conda create -n musetalk python=3.10

进入conda环境

conda activate musetalk

下载项目依赖包

pip install -r requirements.txt

mmlab 封装

pip install --no-cache-dir -U openmim 
mim install mmengine 
mim install "mmcv>=2.0.1" 
mim install "mmdet>=3.1.0" 
mim install "mmpose>=1.1.0" 

注意：这一步可能提示错误，跳过即可

下载 ffmpeg-4.1.4-amd64-static

https://ffmpeg.org/download.html#releases

解压文件ffmpeg-4.1.4-amd64-static

tar -xvf ffmpeg-4.1.4-amd64-static.gz

设置环境变量

vi ~/.bashrc

最后一行填写上

export FFMPEG_PATH=/root/workspace/MuseTalk/musetalk/ffmpeg-4.1.4-amd64-static

使配置生效

source ~/.bashrc

验证安装：验证 ffmpeg 是否正确安装和配置：

$FFMPEG_PATH/ffmpeg -version

(musetalk) [root@iZ0jl0y9289xkrzfhm4p2wZ MuseTalk]# $FFMPEG_PATH/ffmpeg -version
ffmpeg version 4.1.4-static https://johnvansickle.com/ffmpeg/  Copyright (c) 2000-2019 the FFmpeg developers
built with gcc 6.3.0 (Debian 6.3.0-18+deb9u1) 20170516
configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc-6 --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gmp --enable-libgme --enable-gray --enable-libaom --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzvbi --enable-libzimg
libavutil      56. 22.100 / 56. 22.100
libavcodec     58. 35.100 / 58. 35.100
libavformat    58. 20.100 / 58. 20.100
libavdevice    58.  5.100 / 58.  5.100
libavfilter     7. 40.101 /  7. 40.101
libswscale      5.  3.100 /  5.  3.100
libswresample   3.  3.100 /  3.  3.100
libpostproc    55.  3.100 / 55.  3.100

验证ffmpeg是不是安装成功

pip list | grep ffmpeg

(musetalk) [root@iZ0jl0y9289xkrzfhm4p2wZ MuseTalk]# pip list | grep ffmpeg
ffmpeg-python                0.2.0
imageio-ffmpeg               0.5.1
(musetalk) [root@iZ0jl0y9289xkrzfhm4p2wZ MuseTalk]# 

安装正确的ffmpeg-python 库：

pip install ffmpeg-python 

下载需要的模型

链接: https://pan.baidu.com/s/1NxELa1cvtu3aDh3d9sB1Jw?pwd=yptf

提取码: yptf

下载后依次放在models目录下面

(musetalk) [root@iZ0jl0y9289xkrzfhm4p2wZ models]# ls -al
total 3905288
drwxr-xr-x  7 root root       4096 Jun 18 21:46 .
drwxr-xr-x 10 root root       4096 Jun 19 14:43 ..
drwxrwxr-x  2 root root       4096 Jun 18 21:19 dwpose
drwxrwxr-x  2 root root       4096 Jun 18 21:20 face-parse-bisent
drwxrwxr-x  2 root root       4096 Jun 18 21:18 musetalk
drwxrwxr-x  2 root root       4096 Jun 18 21:20 sd-vae-ft-mse
drwxrwxr-x  2 root root       4096 Jun 18 21:20 whisper

models的目录结构为：
./models/
├── musetalk
│ └── musetalk.json
│ └── pytorch_model.bin
├── dwpose
│ └── dw-ll_ucoco_384.pth
├── face-parse-bisent
│ ├── 79999_iter.pth
│ └── resnet18-5c106cde.pth
├── sd-vae-ft-mse
│ ├── config.json
│ └── diffusion_pytorch_model.bin
└── whisper
└── tiny.pt

准备没有声音的视频文件和音频文件

MuseTalk/data/video
MuseTalk/data/audio

编辑配置文件

task_0:
 video_path: "data/video/baichuanxu.mp4"
 audio_path: "data/audio/baichunxu.wav"
 bbox_shift: -7

模型推理

python -m scripts.inference --inference_config configs/inference/test.yaml 

下载视频，到results目录下载生成的视频