soderstrom/scripts/script_download_mp3.py

# -*- coding: utf-8 -*-
"""script_download_mp3.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/12QC9ULnH-_nhwNOO9phzGEt57rSVysM0
"""

# script to download the audio files:
import os
import re
import urllib.request

child_list = ['Joe','The']
for child in child_list:   
    files = os.listdir(f'Soderstrom/{child}/')
    for f in files:
      path = f'https://media.talkbank.org/childes/Eng-NA/Soderstrom/{child}/{re.split("[.]",f)[0]}.mp3'
      urllib.request.urlretrieve(path, f'soderstrom/recordings/raw/{child}/{re.split("[.]",f)[0]}.mp3')