# train model# if you wish to resume from an exists model, uncomment --checkpoint_path and --fastspeech_step#CUDA_VISIBLE_DEVICES=0,1,2,3 \python -u train.py \--batch_size=32 \--epochs=10000 \--lr=0.001 \--save_step=500 \--use_gpu=1 \--use_data_parallel=0 \--data_path='../../dataset/LJSpeech-1.1'\--transtts_path='../transformer_tts/checkpoint'\--transformer_step=160000 \--save_path='./checkpoint'\--log_dir='./log'\--config_path='config/fastspeech.yaml'\#--checkpoint_path='./checkpoint' \#--fastspeech_step=97000 \if[$?-ne 0 ];thenecho"Failed in training!"exit 1fiexit 0
