Mozilla Common Voiceデータ抽出用スクリプト https://note.com/yamimaki_reru/n/n049fb1498365

commonvoice_copy_files.py

#!/usr/bin/env python

# Common Voiceの各mp3を作業用フォルダにコピーするためのスクリプト

import os
import shutil

def main():
    cur_dir = os.path.dirname(__file__)
    dest_dir = 'C:\work\in\cv'
    os.makedirs(dest_dir, exist_ok=True)
    with open(os.path.join(cur_dir, 'to-copy.csv')) as f:
        for line in f:
            line = line.strip()
            if 0 == len(line):
                break
            (dname, fname) = line.split(',')
            shutil.copy(os.path.join(cur_dir, dname, fname),
                    os.path.join(dest_dir, fname))

if __name__ == '__main__':
    main()

commonvoice_get_folders_by_files.py

#!/usr/bin/env python

# Common Voiceの各mp3が入っているフォルダを探しCSVを吐き出すために書いたスクリプト

import os

def main():
    file_dir_map = dict()
    cur_dir = os.path.dirname(__file__)
    with open(os.path.join(cur_dir, 'files.csv')) as f:
        for line in f:
            line = line.strip()
            if 0 == len(line):
                break
            (fname, dname) = line.split(',')
            file_dir_map[fname] = dname
    buf = ''
    with open(os.path.join(cur_dir, 'targets.csv')) as f:
        for line in f:
            line = line.strip()
            if 0 == len(line):
                break
            buf += file_dir_map[line] + "\n"
    with open(os.path.join(cur_dir, 'out-targets.csv'), mode='w') as f:
        f.write(buf)

if __name__ == '__main__':
    main()

commonvoice_list_files.py

#!/usr/bin/env python

# Common Voiceの「ファイル名,ディレクトリ名」のCSVを吐き出すために書いたスクリプト

import os
import math

def main():
    buf = ''
    dir_count = 0
    cur_dir = os.path.dirname(__file__)
    with open(os.path.join(cur_dir, 'files.csv'), mode='w') as f:
        with os.scandir(cur_dir) as dir_it:
            for dir_ent in dir_it:
                if not dir_ent.is_dir():
                    continue
                dir_count += 1
                for file_ent in os.listdir(os.path.join(cur_dir, dir_ent)):
                    buf += f"{file_ent},{dir_ent.name}\n"
                f.write(buf)
                buf = ''

if __name__ == '__main__':
    main()