search

modelscope / 3D-Speaker

A Repository for Single- and Multi-modal Speaker Verification, Speaker Recognition and Speaker Diarization
Apache License 2.0
1.2k stars 101 forks source link

语种识别，怎么将modelscope里的模型对应上 #133

Closed wjyfelicity closed 2 months ago

wjyfelicity commented 2 months ago

用3D-Speaker/egs/3dspeaker/language-identification/run.sh跑出来的模型有三个，CKPT.yaml classifier.ckpt embedding_model.ckpt epoch_counter.ckpt label_encoder.ckpt 而modelscope里的模型只有两个： ihsq19sMbK 而且配置文件也不一样，请问要怎么对应上？

yfchenlucky commented 2 months ago

CKPT.yaml 在推理阶段不需要使用，是模型训练阶段保存的文件。