#!/usr/bin/env bash
#PBS -l select=1:ncpus=16:ngpus=1:mem=64g
#PBS -l walltime=72:00:00
#PBS -q tl-gpu
#PBS -j oe
#PBS -o output.log
#PBS -k oed


# bash /workspace/echoloc/codes/TTS-Framework/examples/streaming_tts/data_prepare/1_data2tsv.py

# PYTHONPATH=/workspace/echoloc/codes/TTS-Framework/ \
# python /workspace/echoloc/codes/TTS-Framework/examples/sft/data_prepare/0_testdata.py 

# PYTHONPATH=/workspace/echoloc/codes/TTS-Framework/ \
# python /workspace/echoloc/codes/TTS-Framework/examples/sft/data_prepare/1_data2tsv.py 


source /workspace/echoloc/.bashrc
conda activate cosyvoice

PYTHONPATH=/workspace/echoloc/codes/TTS-Framework \
python /workspace/echoloc/codes/TTS-Framework/examples/cosyvoice2/data_prepare_tamil/2_tsv2traininfo.py \
  --model_dir /workspace/echoloc/codes/TTS-Framework/pretrained_models/CosyVoice2-0.5B \
  --data_dir /workspace/echoloc/datas/ntu_enzh_speech \
  --out_home /workspace/echoloc/datas/ntu_enzh_speech/sft \
  --splits train test \
  --num_workers 8  2>&1 | tee /workspace/echoloc/datas/ntu_enzh_speech/sft/log.out