def _maybe_split_stm_dataset(extracted_dir, data_set):
# Create stm dir
stm_dir = path.join(extracted_dir, data_set, "stm")
# Obtain stm files
stm_files = glob(path.join(stm_dir, "*.stm"))
# Loop over stm files and split each one
for stm_file in stm_files:
# Parse stm file
stm_segments = parse_stm_file(stm_file)
# Loop over stm_segments and create txt file for each one
for stm_segment in stm_segments:
start_time = stm_segment.start_time
stop_time = stm_segment.stop_time
txt_filename = path.splitext(
path.basename(stm_file))[0] + "-" + str(
start_time) + "-" + str(stop_time) + ".txt"
txt_file = path.join(stm_dir, txt_filename)
# If the txt segment file does not exist create it
if not gfile.Exists(txt_file):
with open(txt_file, "w+") as f:
f.write(stm_segment.transcript)
# Remove stm_file
remove(stm_file)
def _maybe_split_dataset(extracted_dir, data_set):
# Create stm dir
stm_dir = path.join(extracted_dir, data_set, "stm")
# Create wav dir
wav_dir = path.join(extracted_dir, data_set, "wav")
files = []
# Loop over stm files and split corresponding wav
for stm_file in glob(path.join(stm_dir, "*.stm")):
# Parse stm file
stm_segments = parse_stm_file(stm_file)
# Open wav corresponding to stm_file
wav_filename = path.splitext(path.basename(stm_file))[0] + ".wav"
wav_file = path.join(wav_dir, wav_filename)
origAudio = wave.open(wav_file, 'r')
# Loop over stm_segments and split wav_file for each segment
for stm_segment in stm_segments:
# Create wav segment filename
start_time = stm_segment.start_time
stop_time = stm_segment.stop_time
new_wav_filename = path.splitext(
path.basename(stm_file))[0] + "-" + str(
start_time) + "-" + str(stop_time) + ".wav"
new_wav_file = path.join(wav_dir, new_wav_filename)
# If the wav segment filename does not exist create it
if not gfile.Exists(new_wav_file):
_split_wav(origAudio, start_time, stop_time, new_wav_file)
new_wav_filesize = path.getsize(new_wav_file)
files.append(
(new_wav_file, new_wav_filesize, stm_segment.transcript))
# Close origAudio
origAudio.close()
# Remove wav_file
remove(wav_file)
return pandas.DataFrame(
data=files, columns=["wav_filename", "wav_filesize", "transcript"])
def _maybe_split_dataset(extracted_dir, data_set):
# Create stm dir
stm_dir = path.join(extracted_dir, data_set, "stm")
# Create wav dir
wav_dir = path.join(extracted_dir, data_set, "wav")
files = []
# Loop over stm files and split corresponding wav
for stm_file in glob(path.join(stm_dir, "*.stm")):
# Parse stm file
stm_segments = parse_stm_file(stm_file)
# Open wav corresponding to stm_file
wav_filename = path.splitext(path.basename(stm_file))[0] + ".wav"
wav_file = path.join(wav_dir, wav_filename)
origAudio = wave.open(wav_file,'r')
# Loop over stm_segments and split wav_file for each segment
for stm_segment in stm_segments:
# Create wav segment filename
start_time = stm_segment.start_time
stop_time = stm_segment.stop_time
new_wav_filename = path.splitext(path.basename(stm_file))[0] + "-" + str(start_time) + "-" + str(stop_time) + ".wav"
new_wav_file = path.join(wav_dir, new_wav_filename)
# If the wav segment filename does not exist create it
if not gfile.Exists(new_wav_file):
_split_wav(origAudio, start_time, stop_time, new_wav_file)
new_wav_filesize = path.getsize(new_wav_file)
files.append((path.abspath(new_wav_file), new_wav_filesize, stm_segment.transcript))
# Close origAudio
origAudio.close()
return pandas.DataFrame(data=files, columns=["wav_filename", "wav_filesize", "transcript"])
