configs/audio2motion/inference/inference.yaml

output_fps: 25
## appearance and motion feature extractor
appearance_feature_extractor_path:  pretrain_weights/decode/v1/first_stage/base_models/appearance_feature_extractor.pth
motion_extractor_path: pretrain_weights/decode/v1/first_stage/base_models/motion_extractor.pth
## SPADEGenerator
spade_generator_path: pretrain_weights/decode/v1/first_stage/base_models/spade_generator.pth
warping_module_path:  pretrain_weights/decode/v1/first_stage/base_models/warping_module.pth
## stitching retargeting module
stitching_retargeting_module_path:  pretrain_weights/decode/v1/first_stage/retargeting_models/stitching_retargeting_module.pth
#

# audio processer config
audio_model_config: configs/audio2motion/model/audio_processer_config.yaml

# motion processer config
motion_processer_config: configs/audio2motion/model/liveportrait_config.yaml

# motion generator model
motion_models_config: configs/audio2motion/model/config.yaml
use_ref_kp: False
motion_generator_path: pretrain_weights/moda/net-200.pth
need_normalized: True

# other configs
device_id: 0
batch_size: 100

source_max_dim: 1280 # the max dim of height and width of source image or video
source_division: 2 # make sure the height and width of source image or video can be divided by this number
input_height: 256
input_width: 256
source_fps: 25
min_video_length: 50
max_video_length: 500