def cut(input_path, output_file, metadata):
segments = metadata['segments']
segments = [segment_seconds(segment) for segment in segments]
with TempFile('.mp3') as temp_file:
# Open a new temporary file to store audio in between processes
if segments:
# Cut audio into segments and create fade in/out
# We need to use a new temporary file for each
# audio segment
temp_segments = [TempFile('.mp3') for segment in segments]
try:
for index, segment in enumerate(segments):
sox = Transformer()
sox.channels(1)
sox.norm(-24)
sox.trim(*segment)
sox.fade(1, 2, 't')
sox.build(input_path, temp_segments[index].path)
if len(segments) > 1:
# Concatenate all the audio segments back together
# and output to our main temporary file
Combiner().build(
[temp_segment.path for temp_segment in temp_segments],
temp_file.path,
'concatenate',
)
else:
# Only one segment so we don't need to combine anything
subprocess.run(
['cp', temp_segments[0].path, temp_file.path])
except Exception as e:
raise (e)
finally:
# Cleanup temporary segment files even on error
if temp_segments:
for temp_segment in temp_segments:
temp_segment.close()
# Second process: filter, compress and EQ the
# audio in temporary file and output to output_file
sox = Transformer()
sox.highpass(100)
sox.lowpass(10000)
sox.compand(0.005, 0.12, 6, [
(-90, -90),
(-70, -55),
(-50, -35),
(-32, -32),
(-24, -24),
(0, -8),
])
sox.equalizer(3000, 1000, 3)
sox.equalizer(280, 120, 3)
sox.build(temp_file.path, output_file)
def process(x):
file_path, text = x
file_name = os.path.splitext(os.path.basename(file_path))[0]
text = text.lower().strip()
audio_path = os.path.join(audio_clips_path, file_path)
output_wav_path = os.path.join(wav_dir, file_name + '.wav')
tfm = Transformer()
tfm.rate(samplerate=args.sample_rate)
tfm.channels(n_channels=args.n_channels)
tfm.build(input_filepath=audio_path, output_filepath=output_wav_path)
duration = sox.file_info.duration(output_wav_path)
return output_wav_path, duration, text
def process(x):
file_path, text = x
file_name = os.path.splitext(os.path.basename(file_path))[0]
text = text.lower().strip()
audio_path = os.path.join(audio_clips_path, file_path)
if os.path.getsize(audio_path) == 0:
logging.warning(f'Skipping empty audio file {audio_path}')
return '', '', ''
output_wav_path = os.path.join(wav_dir, file_name + '.wav')
if not os.path.exists(output_wav_path):
tfm = Transformer()
tfm.rate(samplerate=args.sample_rate)
tfm.channels(n_channels=args.n_channels)
tfm.build(input_filepath=audio_path, output_filepath=output_wav_path)
duration = sox.file_info.duration(output_wav_path)
return output_wav_path, duration, text
