mrlapuda

I've change the code of inference.py : `import argparse import tqdm import torch import os import utils.inference_util as inference_util import utils.video_util as video_util from utils.common import tensor2img from configs.config import...

key point is ` ``sadtalker_path = { 'audio2pose_yaml_path': audio2pose_yaml_path, 'audio2exp_yaml_path': audio2exp_yaml_path, 'audio2pose_checkpoint': audio2pose_checkpoint, 'audio2exp_checkpoint': audio2exp_checkpoint, 'use_safetensor': False, # 假设您不使用 safetensor，如果使用，请设置为 True 'checkpoint': None, # 如果您使用 safetensor，请提供相应的路径 } audio_to_coeff = Audio2Coeff(sadtalker_path,...