def test_read_write_wav(tmpdir, func, dtype):
path = tmpdir.mkdir('test')
ark = path.join('a.ark').strpath
scp = path.join('a.scp').strpath
# Write as pcm16
array = np.random.randint(0, 10, 10, dtype=dtype)
array2 = np.random.randint(0, 10, 10, dtype=dtype)
d = {'utt': (8000, array), 'utt2': (8000, array2)}
save_ark(ark, d, scp=scp)
d = dict(func(scp))
rate, test = d['utt']
assert rate == 8000
np.testing.assert_array_equal(array, test)
rate, test = d['utt2']
assert rate == 8000
np.testing.assert_array_equal(array2, test)
d = dict(load_ark(ark))
rate, test = d['utt']
assert rate == 8000
np.testing.assert_array_equal(array, test)
rate, test = d['utt2']
assert rate == 8000
np.testing.assert_array_equal(array2, test)
def __iter__(self):
if self.scp:
while True:
try:
k, v = next(self.generator)
except StopIteration:
break
except Exception:
if self.permissive:
# Stop if error happen
break
else:
raise
yield k, v
else:
with self.file as f:
it = load_ark(f)
while True:
try:
k, v = next(it)
except StopIteration:
break
except Exception:
if self.permissive:
# Stop if error happen
break
else:
raise
yield k, v
self.closed = True
def test_wavark_stream(tmpdir, dtype, write_function):
path = tmpdir.mkdir("test")
ark = path.join("a.ark").strpath
# Write as pcm16
array = np.random.randint(-1000, 1000, 100).astype(np.double) / abs(
np.iinfo(np.int16).min)
array2 = np.random.randint(-1000, 1000, 100).astype(np.double) / abs(
np.iinfo(np.int16).min)
if write_function == "numpy":
d = {"utt": array, "utt2": array2}
else:
d = {"utt": (8000, array), "utt2": (8000, array2)}
save_ark(ark, d, write_function=write_function)
with open_like_kaldi("cat {}|".format(ark), "rb") as f:
d = dict(load_ark(f))
if write_function == "numpy":
test = d["utt"]
else:
rate, test = d["utt"]
assert rate == 8000
np.testing.assert_allclose(array, test)
if write_function == "numpy":
test = d["utt2"]
else:
rate, test = d["utt2"]
assert rate == 8000
np.testing.assert_allclose(array2, test)
def test_read_write_wav(tmpdir, func, dtype):
path = tmpdir.mkdir("test")
ark = path.join("a.ark").strpath
scp = path.join("a.scp").strpath
# Write as pcm16
array = np.random.randint(0, 10, 10, dtype=dtype)
array2 = np.random.randint(0, 10, 10, dtype=dtype)
d = {"utt": (8000, array), "utt2": (8000, array2)}
save_ark(ark, d, scp=scp)
d = dict(func(scp))
rate, test = d["utt"]
assert rate == 8000
np.testing.assert_array_equal(array, test)
rate, test = d["utt2"]
assert rate == 8000
np.testing.assert_array_equal(array2, test)
d = dict(load_ark(ark))
rate, test = d["utt"]
assert rate == 8000
np.testing.assert_array_equal(array, test)
rate, test = d["utt2"]
assert rate == 8000
np.testing.assert_array_equal(array2, test)
def test_write_helper_scp_ark(tmpdir):
path = tmpdir.strpath
d = {"foo": numpy.random.randn(10, 10), "bar": numpy.random.randn(10, 10)}
with WriteHelper("scp,f,ark:{p}/out.scp,{p}/out.ark".format(p=path)) as w:
for k, v in d.items():
w(k, v)
from_ark = dict(load_ark("{p}/out.ark".format(p=path)))
from_scp = load_scp("{p}/out.scp".format(p=path))
_compare(from_ark, d)
_compare(from_scp, d)
def test_write_helper(tmpdir):
path = tmpdir.strpath
d = {'foo': numpy.random.randn(10, 10), 'bar': numpy.random.randn(10, 10)}
with WriteHelper('ark,f,scp:{p}/out.ark,{p}/out.scp'.format(p=path)) as w:
for k, v in d.items():
w(k, v)
from_ark = dict(load_ark('{p}/out.ark'.format(p=path)))
from_scp = load_scp('{p}/out.scp'.format(p=path))
_compare(from_ark, d)
_compare(from_scp, d)
def test_read_write(tmpdir, func, dtype, write_function):
path = tmpdir.mkdir("test")
ark = path.join("a.ark").strpath
scp = path.join("a.scp").strpath
# Write as pcm16
array = np.random.randint(-1000, 1000, 100).astype(np.double) / float(
abs(np.iinfo(np.int16).min))
array2 = np.random.randint(-1000, 1000, 100).astype(np.double) / abs(
np.iinfo(np.int16).min)
if write_function == "numpy":
d = {"utt": array, "utt2": array2}
else:
d = {"utt": (8000, array), "utt2": (8000, array2)}
save_ark(ark, d, scp=scp, write_function=write_function)
d = dict(func(scp))
if write_function == "numpy":
test = d["utt"]
else:
rate, test = d["utt"]
assert rate == 8000
np.testing.assert_allclose(array, test)
if write_function == "numpy":
test = d["utt2"]
else:
rate, test = d["utt2"]
assert rate == 8000
np.testing.assert_allclose(array2, test)
d = dict(load_ark(ark))
if write_function == "numpy":
test = d["utt"]
else:
rate, test = d["utt"]
assert rate == 8000
np.testing.assert_allclose(array, test)
if write_function == "numpy":
test = d["utt2"]
else:
rate, test = d["utt2"]
assert rate == 8000
np.testing.assert_allclose(array2, test)
def test_wavark_stream(tmpdir, dtype):
path = tmpdir.mkdir('test')
ark = path.join('a.ark').strpath
# Write as pcm16
array = np.random.randint(0, 10, 10, dtype=dtype)
array2 = np.random.randint(0, 10, 10, dtype=dtype)
d = {'utt': (8000, array), 'utt2': (8000, array2)}
save_ark(ark, d)
with open_like_kaldi('cat {}|'.format(ark), 'rb') as f:
d = dict(load_ark(f))
rate, test = d['utt']
assert rate == 8000
np.testing.assert_array_equal(array, test)
rate, test = d['utt2']
assert rate == 8000
np.testing.assert_array_equal(array2, test)
def test_wavark_stream(tmpdir, dtype):
path = tmpdir.mkdir("test")
ark = path.join("a.ark").strpath
# Write as pcm16
array = np.random.randint(0, 10, 10, dtype=dtype)
array2 = np.random.randint(0, 10, 10, dtype=dtype)
d = {"utt": (8000, array), "utt2": (8000, array2)}
save_ark(ark, d)
with open_like_kaldi("cat {}|".format(ark), "rb") as f:
d = dict(load_ark(f))
rate, test = d["utt"]
assert rate == 8000
np.testing.assert_array_equal(array, test)
rate, test = d["utt2"]
assert rate == 8000
np.testing.assert_array_equal(array2, test)
