def main():
""" Entry point """
if len(sys.argv) < 3:
usage()
return
file_path = sys.argv[1]
save_path = sys.argv[2]
if not gf.can_run_c_extension():
print "[WARN] Unable to load Python C Extensions"
print "[WARN] Running the slower pure Python code"
print "[WARN] See the README file for directions to compile the Python C Extensions"
audiofile = AudioFile(file_path)
audiofile.load_data()
audiofile.extract_mfcc()
audiofile.clear_data()
numpy.savetxt(save_path, audiofile.audio_mfcc)
print "[INFO] MFCCs saved to %s" % (save_path)
def _cut_head_tail(self, audio_file_path):
"""
Set the audio file head or tail,
suitably cutting the audio file on disk,
and setting the corresponding parameters in the task configuration.
Return a success bool flag
"""
self._log("Setting head and/or tail")
try:
configuration = self.task.configuration
head_length = configuration.is_audio_file_head_length
process_length = configuration.is_audio_file_process_length
detect_head_min = configuration.is_audio_file_detect_head_min
detect_head_max = configuration.is_audio_file_detect_head_max
detect_tail_min = configuration.is_audio_file_detect_tail_min
detect_tail_max = configuration.is_audio_file_detect_tail_max
# explicit head or process?
explicit = (head_length is not None) or (process_length
is not None)
# at least one detect parameter?
detect = ((detect_head_min is not None)
or (detect_head_max is not None)
or (detect_tail_min is not None)
or (detect_tail_max is not None))
if explicit or detect:
# we need to load the audio data
audio_file = AudioFile(audio_file_path, logger=self.logger)
audio_file.load_data()
if explicit:
self._log("Explicit head or process")
else:
self._log(
"No explicit head or process => detecting head/tail")
head = 0.0
if (detect_head_min is not None) or (detect_head_max
is not None):
self._log("Detecting head...")
detect_head_min = gf.safe_float(
detect_head_min, gc.SD_MIN_HEAD_LENGTH)
detect_head_max = gf.safe_float(
detect_head_max, gc.SD_MAX_HEAD_LENGTH)
self._log(["detect_head_min is %.3f", detect_head_min])
self._log(["detect_head_max is %.3f", detect_head_max])
sd = SD(audio_file,
self.task.text_file,
logger=self.logger)
head = sd.detect_head(detect_head_min, detect_head_max)
self._log(["Detected head: %.3f", head])
tail = 0.0
if (detect_tail_min is not None) or (detect_tail_max
is not None):
self._log("Detecting tail...")
detect_tail_max = gf.safe_float(
detect_tail_max, gc.SD_MAX_TAIL_LENGTH)
detect_tail_min = gf.safe_float(
detect_tail_min, gc.SD_MIN_TAIL_LENGTH)
self._log(["detect_tail_min is %.3f", detect_tail_min])
self._log(["detect_tail_max is %.3f", detect_tail_max])
sd = SD(audio_file,
self.task.text_file,
logger=self.logger)
tail = sd.detect_tail(detect_tail_min, detect_tail_max)
self._log(["Detected tail: %.3f", tail])
# sanity check
head_length = max(0, head)
process_length = max(0,
audio_file.audio_length - tail - head)
# we need to set these values
# in the config object for later use
self.task.configuration.is_audio_file_head_length = head_length
self.task.configuration.is_audio_file_process_length = process_length
self._log(["Set head_length: %.3f", head_length])
self._log(["Set process_length: %.3f", process_length])
if head_length is not None:
# in case we are reading from config object
head_length = float(head_length)
if process_length is not None:
# in case we are reading from config object
process_length = float(process_length)
# note that str() is necessary, as one might be None
self._log(
["is_audio_file_head_length is %s",
str(head_length)])
self._log([
"is_audio_file_process_length is %s",
str(process_length)
])
self._log("Trimming audio data...")
audio_file.trim(head_length, process_length)
self._log("Trimming audio data... done")
self._log("Writing audio file...")
audio_file.write(audio_file_path)
self._log("Writing audio file... done")
audio_file.clear_data()
else:
# nothing to do
self._log("No explicit head/process or detect head/tail")
self._log("Setting head and/or tail: succeeded")
return True
except Exception as e:
self._log("Setting head and/or tail: failed")
self._log(["Message: %s", str(e)])
return False
def _cut_head_tail(self, audio_file_path):
"""
Set the audio file head or tail,
suitably cutting the audio file on disk,
and setting the corresponding parameters in the task configuration.
Return a success bool flag
"""
self._log("Setting head and/or tail")
try:
configuration = self.task.configuration
head_length = configuration.is_audio_file_head_length
process_length = configuration.is_audio_file_process_length
detect_head_min = configuration.is_audio_file_detect_head_min
detect_head_max = configuration.is_audio_file_detect_head_max
detect_tail_min = configuration.is_audio_file_detect_tail_min
detect_tail_max = configuration.is_audio_file_detect_tail_max
# explicit head or process?
explicit = (head_length is not None) or (process_length is not None)
# at least one detect parameter?
detect = (
(detect_head_min is not None) or
(detect_head_max is not None) or
(detect_tail_min is not None) or
(detect_tail_max is not None)
)
if explicit or detect:
# we need to load the audio data
audio_file = AudioFile(audio_file_path, logger=self.logger)
audio_file.load_data()
if explicit:
self._log("Explicit head or process")
else:
self._log("No explicit head or process => detecting head/tail")
head = 0.0
if (detect_head_min is not None) or (detect_head_max is not None):
self._log("Detecting head...")
detect_head_min = gf.safe_float(detect_head_min, gc.SD_MIN_HEAD_LENGTH)
detect_head_max = gf.safe_float(detect_head_max, gc.SD_MAX_HEAD_LENGTH)
self._log(["detect_head_min is %.3f", detect_head_min])
self._log(["detect_head_max is %.3f", detect_head_max])
sd = SD(audio_file, self.task.text_file, logger=self.logger)
head = sd.detect_head(detect_head_min, detect_head_max)
self._log(["Detected head: %.3f", head])
tail = 0.0
if (detect_tail_min is not None) or (detect_tail_max is not None):
self._log("Detecting tail...")
detect_tail_max = gf.safe_float(detect_tail_max, gc.SD_MAX_TAIL_LENGTH)
detect_tail_min = gf.safe_float(detect_tail_min, gc.SD_MIN_TAIL_LENGTH)
self._log(["detect_tail_min is %.3f", detect_tail_min])
self._log(["detect_tail_max is %.3f", detect_tail_max])
sd = SD(audio_file, self.task.text_file, logger=self.logger)
tail = sd.detect_tail(detect_tail_min, detect_tail_max)
self._log(["Detected tail: %.3f", tail])
# sanity check
head_length = max(0, head)
process_length = max(0, audio_file.audio_length - tail - head)
# we need to set these values
# in the config object for later use
self.task.configuration.is_audio_file_head_length = head_length
self.task.configuration.is_audio_file_process_length = process_length
self._log(["Set head_length: %.3f", head_length])
self._log(["Set process_length: %.3f", process_length])
if head_length is not None:
# in case we are reading from config object
head_length = float(head_length)
if process_length is not None:
# in case we are reading from config object
process_length = float(process_length)
# note that str() is necessary, as one might be None
self._log(["is_audio_file_head_length is %s", str(head_length)])
self._log(["is_audio_file_process_length is %s", str(process_length)])
self._log("Trimming audio data...")
audio_file.trim(head_length, process_length)
self._log("Trimming audio data... done")
self._log("Writing audio file...")
audio_file.write(audio_file_path)
self._log("Writing audio file... done")
audio_file.clear_data()
else:
# nothing to do
self._log("No explicit head/process or detect head/tail")
self._log("Setting head and/or tail: succeeded")
return True
except Exception as e:
self._log("Setting head and/or tail: failed")
self._log(["Message: %s", str(e)])
return False
