def play_collection():
augmentations = parse_augmentations(CLI_ARGS.augment)
if any(not isinstance(a, SampleAugmentation) for a in augmentations):
print("Warning: Some of the augmentations cannot be simulated by this command.")
samples = get_samples_in_play_order()
samples = apply_sample_augmentations(samples,
audio_type=AUDIO_TYPE_PCM,
augmentations=augmentations,
process_ahead=0,
clock=CLI_ARGS.clock)
for sample in samples:
if not CLI_ARGS.quiet:
print('Sample "{}"'.format(sample.sample_id), file=sys.stderr)
if isinstance(sample, LabeledSample):
print(' "{}"'.format(sample.transcript), file=sys.stderr)
if CLI_ARGS.pipe:
sample.change_audio_type(AUDIO_TYPE_WAV)
sys.stdout.buffer.write(sample.audio.getvalue())
return
wave_obj = simpleaudio.WaveObject(sample.audio,
sample.audio_format.channels,
sample.audio_format.width,
sample.audio_format.rate)
play_obj = wave_obj.play()
play_obj.wait_done()
def build_sdb():
audio_type = AUDIO_TYPE_LOOKUP[CLI_ARGS.audio_type]
augmentations = parse_augmentations(CLI_ARGS.augment)
if any(not isinstance(a, SampleAugmentation) for a in augmentations):
print(
"Warning: Some of the augmentations cannot be applied by this command."
)
with DirectSDBWriter(CLI_ARGS.target,
audio_type=audio_type,
labeled=not CLI_ARGS.unlabeled) as sdb_writer:
samples = samples_from_sources(CLI_ARGS.sources,
labeled=not CLI_ARGS.unlabeled)
num_samples = len(samples)
if augmentations:
samples = apply_sample_augmentations(samples,
audio_type=AUDIO_TYPE_PCM,
augmentations=augmentations)
bar = progressbar.ProgressBar(max_value=num_samples,
widgets=SIMPLE_BAR)
for sample in bar(
change_audio_types(samples,
audio_type=audio_type,
bitrate=CLI_ARGS.bitrate,
processes=CLI_ARGS.workers)):
sdb_writer.add(sample)
def build_data_set():
audio_type = AUDIO_TYPE_LOOKUP[CLI_ARGS.audio_type]
augmentations = parse_augmentations(CLI_ARGS.augment)
if any(not isinstance(a, SampleAugmentation) for a in augmentations):
print(
'Warning: Some of the specified augmentations will not get applied, as this tool only supports '
'overlay, codec, reverb, resample and volume.')
extension = Path(CLI_ARGS.target).suffix.lower()
labeled = not CLI_ARGS.unlabeled
if extension == '.csv':
writer = CSVWriter(CLI_ARGS.target,
absolute_paths=CLI_ARGS.absolute_paths,
labeled=labeled)
elif extension == '.sdb':
writer = DirectSDBWriter(CLI_ARGS.target,
audio_type=audio_type,
labeled=labeled)
elif extension == '.tar':
writer = TarWriter(CLI_ARGS.target,
labeled=labeled,
gz=False,
include=CLI_ARGS.include)
elif extension == '.tgz' or CLI_ARGS.target.lower().endswith('.tar.gz'):
writer = TarWriter(CLI_ARGS.target,
labeled=labeled,
gz=True,
include=CLI_ARGS.include)
else:
print(
'Unknown extension of target file - has to be either .csv, .sdb, .tar, .tar.gz or .tgz'
)
sys.exit(1)
with writer:
samples = samples_from_sources(CLI_ARGS.sources,
labeled=not CLI_ARGS.unlabeled)
num_samples = len(samples)
if augmentations:
samples = apply_sample_augmentations(samples,
audio_type=AUDIO_TYPE_PCM,
augmentations=augmentations)
bar = progressbar.ProgressBar(max_value=num_samples,
widgets=SIMPLE_BAR)
for sample in bar(
change_audio_types(samples,
audio_type=audio_type,
bitrate=CLI_ARGS.bitrate,
processes=CLI_ARGS.workers)):
writer.add(sample)