def _generate_index(self):
target_file = open(os.path.join(self._dir, self._name), "w")
self._make_dirs()
#only get file names
files = [_ for _ in os.walk(os.path.join(self._dir, "in"))][0][-1]
for file in files:
#resample to 16kHz and place resampled in out dir
sox("sox {0} {1} rate 16000".format(
os.path.join(self._dir, "in", file),
os.path.join(self._dir, "out", file)).split())
#copy each resampled file and overwrite the high sample rate file
shutil.copy(os.path.join(self._dir, "out", file),
os.path.join(self._dir, "in", file))
#find the frequencies as numbers in the file name. most of these will be '##.' if so remove '.'
note = file[12:15]
if note[-1] == '.':
note = note[:-1]
#this means output freq isn't in dataset
if int(note) > 96:
continue
#octaves are 12 semitones apart. thus have output be 12 higher
print(file[:12] + "{0}.wav,".format(int(note)) + file[:12] +
"{0}.wav".format(int(note) + 12),
file=target_file)
target_file.close()
def splitFile(fileName):
args = [
fileName, 'outfile.wav', 'silence', '1', '0.1', '0.1', '1', '0.1',
'0.1', ': newfile', ': restart'
]
# Calls the silence argument to `sox`
# - args is a list of arguments passed
# - Will create set of wav files in working directory sequentially
# named outfile (outfile001, outfile002, ...)
# - Starts splitting audio when .1 seconds of 1% of volume is detected
# - Stops splitting audio after .1 seconds of silence is detected
# - Loops through entire file, so can do whole captcha message in one go!
sx.sox(args)
def test_sox_fail_bad_args(self):
args = ['-asdf']
expected_status = 1
actual_status, actual_out, acutal_err = core.sox(args)
self.assertEqual(expected_status, actual_status)
self.assertNotEqual('', actual_out)
self.assertNotEqual('', acutal_err)
def test_sox_fail_bad_args(self):
args = ['-asdf']
expected_status = 1
actual_status, actual_out, acutal_err = core.sox(args)
self.assertEqual(expected_status, actual_status)
self.assertNotEqual('', actual_out)
self.assertNotEqual('', acutal_err)
def test_sox_fail_corrupt_file(self):
args = [INPUT_FILE_CORRUPT, OUTPUT_FILE]
expected_status = 2
expected_out = ''
actual_status, actual_out, acutal_err = core.sox(args)
self.assertEqual(expected_status, actual_status)
self.assertEqual(expected_out, actual_out)
self.assertNotEqual('', acutal_err)
def test_sox_fail_bad_ext(self):
args = ['input.wav', 'output.xyz']
expected_status = 2
expected_out = ''
actual_status, actual_out, acutal_err = core.sox(args)
self.assertEqual(expected_status, actual_status)
self.assertEqual(expected_out, actual_out)
self.assertNotEqual('', acutal_err)
def test_sox_fail_bad_files(self):
args = ['asdf.wav', 'flululu.wav']
expected_status = 2
expected_out = ''
actual_status, actual_out, acutal_err = core.sox(args)
self.assertEqual(expected_status, actual_status)
self.assertEqual(expected_out, actual_out)
self.assertNotEqual('', acutal_err)
def test_sox_fail_corrupt_file(self):
args = [INPUT_FILE_CORRUPT, OUTPUT_FILE]
expected_status = 2
expected_out = ''
actual_status, actual_out, acutal_err = core.sox(args)
self.assertEqual(expected_status, actual_status)
self.assertEqual(expected_out, actual_out)
self.assertNotEqual('', acutal_err)
def test_sox_fail_bad_ext(self):
args = ['input.wav', 'output.xyz']
expected_status = 2
expected_out = ''
actual_status, actual_out, acutal_err = core.sox(args)
self.assertEqual(expected_status, actual_status)
self.assertEqual(expected_out, actual_out)
self.assertNotEqual('', acutal_err)
def test_sox_fail_bad_files(self):
args = ['asdf.wav', 'flululu.wav']
expected_status = 2
expected_out = ''
actual_status, actual_out, acutal_err = core.sox(args)
self.assertEqual(expected_status, actual_status)
self.assertEqual(expected_out, actual_out)
self.assertNotEqual('', acutal_err)
def test_src_array_invalid(self):
args = ['input.wav', 'output.xyz']
arr = 'not a numpy array'
expected_status = 1
actual_status, _, _ = core.sox(args, arr)
self.assertEqual(expected_status, actual_status)
def test_base_case2(self):
args = [INPUT_FILE, OUTPUT_FILE]
expected = (0, '', '')
actual = core.sox(args)
self.assertEqual(expected, actual)
def test_base_case_pathlib(self):
args = ['sox', Path(INPUT_FILE), Path(OUTPUT_FILE)]
expected = (0, '', '')
actual = core.sox(args)
self.assertEqual(expected, actual)
os.stat(os.path.join(Base_Dir, str(counter)))
except:
os.mkdir(os.path.join(Base_Dir, str(counter)))
counter += 1
counter = 0
for file in glob.glob("*.wav"):
if os.path.getsize(file) > 100:
args = [file, 'out.wav', 'silence',
'1', '0.1', '0.1',
'1', '0.1', '0.1',
': newfile', ': restart']
sx.sox(args)
pos = 0
for c in os.path.splitext(file)[0]:
if pos < 9:
out = "out00" + str(pos + 1) + ".wav"
else:
out = "out0" + str(pos + 1) + ".wav"
new = str(counter) + ".wav"
oldLoc = os.path.join(Base_Dir, new)
newLoc = os.path.join(os.path.join(Base_Dir, str(c)), new)
print "Old - " + out
print "New - " + new
os.rename(out, new)
shutil.move(oldLoc, newLoc)
pos += 1
def test_base_case2(self):
args = [INPUT_FILE, OUTPUT_FILE]
expected = (0, '', '')
actual = core.sox(args)
self.assertEqual(expected, actual)