ds_name: TH1KH_512
raw_data_dir: /mnt/bn/sa-ag-data/yezhenhui/datasets/raw/TH1KH_512
binary_data_dir: data/binary/TH1KH_512
# binary_data_dir: /dev/shm/TH1KH
process_id: 0 # rank id when pre-processing dataset
total_process: 1 # number of ranks when pre-processing dataset
split_seed: 999 # random seed that split chunks during pre-processing dataset

max_sentences_per_batch: 1024
max_tokens_per_batch: 200000

load_db_to_memory: false

num_workers: 4
use_kv_dataset: true

binarization_args:
  with_hubert: false
  with_mel: false
  with_coeff: true