process pavoque labels

pavoque.py

import os
import glob
from shutil import copyfile

base_dir = '/Tmp/sotelo/data/german/raw'
text_dir = os.path.join(base_dir, 'text')
wav_dir = os.path.join(base_dir, 'wav/')

audio_files = [
    os.path.join(root, name)
    for root, dirs, files in os.walk(wav_dir)
    for name in files
    if name.endswith((".wav"))]
audio_files = sorted(audio_files)

text_files = [
    os.path.join(root, name)
    for root, dirs, files in os.walk(text_dir)
    for name in files
    if name.endswith((".txt"))]
text_files = sorted(text_files)

audio_names = [x.strip().split('/')[-1].split('.')[0] for x in audio_files]
text_names = [x.strip().split('/')[-1].split('.')[0] for x in text_files]

audio_names = [x.split('-') for x in audio_names]

audio_names, audio_emotions = zip(*audio_names)

all_text = {}
for text_file, text_name in zip(text_files, text_names):
    with open(text_file, 'r') as f:
        text_data = f.readlines()

    assert len(text_data) == 1

    all_text[text_name] = text_data[0]

proc_txt = []
for audio_file in audio_files:
    audio_name = audio_file.strip().split('/')[-1].split('.')[0]
    audio_code, audio_emotion = audio_name.split('-')
    text = all_text[audio_code]

    proc_txt.append('( ' + audio_name + ' "' + text + '" )\n')

with open(os.path.join(base_dir, 'utts.data'), 'w') as f:
    f.writelines(proc_txt)