import os
import soundfile as sf
import random
from tqdm import tqdm
import glob


train_dir = "/data4/liandong/datasets/Noises/DNS_16k/DNS-Challenge-NOISE-16k"
train_scp_file  = "/data4/liandong/datasets/Noises/DNS_16k/DNS_16k.scp"

train_file_list = glob.glob(f'{train_dir}/*.wav')

shuffle_flag = True
if not shuffle_flag:
    train_file_list.sort()
else:
    random.shuffle(train_file_list)

train_file_num = len(train_file_list)
print('The number of files: {}'.format(train_file_num))
train_scp = open(train_scp_file, 'w')
for file_idx in tqdm(range(train_file_num)):
    train_path = train_file_list[file_idx]
    sig, _ = sf.read(train_path)
    if (sig ** 2).sum() > 1e-5:
        train_scp.write(train_path + '\n')
    else:
        print(f"The filename {os.path.split(train_path)[-1]} should be filtered.")
        os.remove(train_path)

print('Write the scp...')
train_scp.close()