Skip to content

Commit

Permalink
Use f-strings.
Browse files Browse the repository at this point in the history 
Reformatted using ruff.
  • Loading branch information
@hendriks73
hendriks73 committed Oct 11, 2024
1 parent 0ee0bc6 commit 0808253
Show file tree
Hide file tree
Showing 7 changed files with 471 additions and 354 deletions.
1 change: 1 addition & 0 deletions tempocnn/__init__.py
View file
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
"""
tempo-cnn is a simple package that allows estimation of musical tempo.
"""

import logging
import tensorflow as tf

Expand Down
145 changes: 88 additions & 57 deletions tempocnn/classifier.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,9 +6,9 @@
from urllib.error import HTTPError

import numpy as np
from tensorflow.python.keras.models import load_model
from tensorflow.keras.models import load_model

logger = logging.getLogger('tempocnn.classifier')
logger = logging.getLogger("tempocnn.classifier")


def std_normalizer(data):
Expand All @@ -23,8 +23,8 @@ def std_normalizer(data):
data = data.astype(np.float64)
mean = np.mean(data)
std = np.std(data)
if std != 0.:
data = (data-mean) / std
if std != 0.0:
data = (data - mean) / std
return data.astype(np.float16)


Expand All @@ -46,44 +46,49 @@ class TempoClassifier:
Classifier that can estimate musical tempo in different formats.
"""

def __init__(self, model_name='fcn'):
def __init__(self, model_name="fcn"):
"""
Initializes this classifier with a Keras model.
:param model_name: model name from sub-package models. E.g. 'fcn', 'cnn', or 'ismir2018'
"""
if 'fma' in model_name:
if "fma" in model_name:
# fma model uses log BPM scale
factor = 256. / np.log(10)
factor = 256.0 / np.log(10)
self.to_bpm = lambda index: np.exp((index + 435) / factor)
else:
self.to_bpm = lambda index: index + 30

# match alias for dt_maz_v fold 0.
if model_name == 'mazurka':
model_name = 'dt_maz_v_fold0'
if model_name == "mazurka":
model_name = "dt_maz_v_fold0"
# match aliases for specific deep/shallow models
elif model_name == 'deeptemp':
model_name = 'deeptemp_k16'
elif model_name == 'shallowtemp':
model_name = 'shallowtemp_k6'
elif model_name == 'deepsquare':
model_name = 'deepsquare_k16'
elif model_name == "deeptemp":
model_name = "deeptemp_k16"
elif model_name == "shallowtemp":
model_name = "shallowtemp_k6"
elif model_name == "deepsquare":
model_name = "deepsquare_k16"
self.model_name = model_name

# mazurka and deeptemp/shallowtempo models use a different kind of normalization
self.normalize = std_normalizer if 'dt_maz' in self.model_name \
or 'deeptemp' in self.model_name \
or 'deepsquare' in self.model_name \
or 'shallowtemp' in self.model_name \
self.normalize = (
std_normalizer
if "dt_maz" in self.model_name
or "deeptemp" in self.model_name
or "deepsquare" in self.model_name
or "shallowtemp" in self.model_name
else max_normalizer
)

resource = _to_model_resource(model_name)
try:
file = _extract_from_package(resource)
except Exception as e:
print('Failed to find a model named \'{}\'. Please check the model name.'.format(model_name),
file=sys.stderr)
print(
f"Failed to find a model named '{model_name}'. Please check the model name.",
file=sys.stderr,
)
raise e
logger.debug(f"Loading model {model_name} from {file}")
self.model = load_model(file)
Expand All @@ -96,10 +101,18 @@ def estimate(self, data):
:param data: features
:return: tempo probability distribution
"""
assert len(data.shape) == 4, 'Input data must be four dimensional. Actual shape was ' + str(data.shape)
assert data.shape[1] == 40, 'Second dim of data must be 40. Actual shape was ' + str(data.shape)
assert data.shape[2] == 256, 'Third dim of data must be 256. Actual shape was ' + str(data.shape)
assert data.shape[3] == 1, 'Fourth dim of data must be 1. Actual shape was ' + str(data.shape)
assert len(data.shape) == 4, (
"Input data must be four dimensional. Actual shape was " + str(data.shape)
)
assert data.shape[1] == 40, (
"Second dim of data must be 40. Actual shape was " + str(data.shape)
)
assert data.shape[2] == 256, (
"Third dim of data must be 256. Actual shape was " + str(data.shape)
)
assert data.shape[3] == 1, (
"Fourth dim of data must be 1. Actual shape was " + str(data.shape)
)
norm_data = self.normalize(data)
return self.model.predict(norm_data, norm_data.shape[0])

Expand All @@ -118,8 +131,8 @@ def quad_interpol_argmax(y, x=None):
return x, y[x]
z = np.polyfit([x - 1, x, x + 1], [y[x - 1], y[x], y[x + 1]], 2)
# find (float) x value for max
argmax = -z[1] / (2. * z[0])
height = z[2] - (z[1] ** 2.) / (4. * z[0])
argmax = -z[1] / (2.0 * z[0])
height = z[2] - (z[1] ** 2.0) / (4.0 * z[0])
return argmax, height

def estimate_tempo(self, data, interpolate=False):
Expand Down Expand Up @@ -158,10 +171,13 @@ def find_index_peaks(distribution):
height = distribution[index]
start = max(index - 5, 0)
length = min(11, distribution.shape[0] - start)
m = np.max(distribution[start:start + length])
m = np.max(distribution[start : start + length])
if height == m and index > last_index + 5:
if interpolate:
interpolated_index, interpolated_height = self.quad_interpol_argmax(distribution, x=index)
(
interpolated_index,
interpolated_height,
) = self.quad_interpol_argmax(distribution, x=index)
p.append((interpolated_index, interpolated_height))
else:
p.append((index, height))
Expand All @@ -173,19 +189,19 @@ def find_index_peaks(distribution):
peaks = find_index_peaks(averaged_prediction)

if len(peaks) == 0:
s1 = 1.
t1 = 0.
t2 = 0.
s1 = 1.0
t1 = 0.0
t2 = 0.0
elif len(peaks) == 1:
bpm = self.to_bpm(peaks[0][0])
if bpm > 120:
alt = bpm/2
s1 = 0.
alt = bpm / 2
s1 = 0.0
t1 = alt
t2 = bpm
else:
alt = bpm*2
s1 = 1.
alt = bpm * 2
s1 = 1.0
t1 = bpm
t2 = alt
else:
Expand All @@ -194,11 +210,11 @@ def find_index_peaks(distribution):
alt = self.to_bpm(peaks[1][0])
alt_height = peaks[1][1]
if bpm < alt:
s1 = bpm_height / (bpm_height+alt_height)
s1 = bpm_height / (bpm_height + alt_height)
t1 = bpm
t2 = alt
else:
s1 = alt_height / (bpm_height+alt_height)
s1 = alt_height / (bpm_height + alt_height)
t1 = alt
t2 = bpm
return t1, t2, s1
Expand All @@ -209,7 +225,7 @@ class MeterClassifier:
Classifier that can estimate musical meter
"""

def __init__(self, model_name='fcn'):
def __init__(self, model_name="fcn"):
"""
Initializes this classifier with a Keras model.
Expand All @@ -218,17 +234,22 @@ def __init__(self, model_name='fcn'):
self._to_meter = lambda index: index + 2
self.model_name = model_name
# mazurka and deeptemp/shallowtempo models use a different kind of normalization
self.normalize = std_normalizer if 'dt_maz_v' in self.model_name \
or 'deeptemp' in self.model_name \
or 'deepsquare' in self.model_name \
or 'shallowtemp' in self.model_name \
self.normalize = (
std_normalizer
if "dt_maz_v" in self.model_name
or "deeptemp" in self.model_name
or "deepsquare" in self.model_name
or "shallowtemp" in self.model_name
else max_normalizer
)
resource = _to_model_resource(model_name)
try:
file = _extract_from_package(resource)
except Exception as e:
print('Failed to find a model named \'{}\'. Please check the model name.'.format(model_name),
file=sys.stderr)
print(
f"Failed to find a model named '{model_name}'. Please check the model name.",
file=sys.stderr,
)
raise e
self.model = load_model(file)

Expand All @@ -240,10 +261,18 @@ def estimate(self, data):
:param data: features
:return: meter probability distribution
"""
assert len(data.shape) == 4, 'Input data must be four dimensional. Actual shape was ' + str(data.shape)
assert data.shape[1] == 40, 'Second dim of data must be 40. Actual shape was ' + str(data.shape)
assert data.shape[2] == 512, 'Third dim of data must be 512. Actual shape was ' + str(data.shape)
assert data.shape[3] == 1, 'Fourth dim of data must be 1. Actual shape was ' + str(data.shape)
assert len(data.shape) == 4, (
"Input data must be four dimensional. Actual shape was " + str(data.shape)
)
assert data.shape[1] == 40, (
"Second dim of data must be 40. Actual shape was " + str(data.shape)
)
assert data.shape[2] == 512, (
"Third dim of data must be 512. Actual shape was " + str(data.shape)
)
assert data.shape[3] == 1, (
"Fourth dim of data must be 1. Actual shape was " + str(data.shape)
)
norm_data = self.normalize(data)
return self.model.predict(norm_data, norm_data.shape[0])

Expand All @@ -262,28 +291,28 @@ def estimate_meter(self, data):

def _to_model_resource(model_name):
file = model_name
if not model_name.endswith('.h5'):
file = file + '.h5'
if not file.startswith('models/'):
file = 'models/' + file
if not model_name.endswith(".h5"):
file = file + ".h5"
if not file.startswith("models/"):
file = "models/" + file
return file


def _extract_from_package(resource):
# check local cache
cache_path = Path(Path.home(), '.tempocnn', resource)
cache_path = Path(Path.home(), ".tempocnn", resource)
if cache_path.exists():
return str(cache_path)

# ensure cache path exists
cache_path.parent.mkdir(parents=True, exist_ok=True)

data = pkgutil.get_data('tempocnn', resource)
data = pkgutil.get_data("tempocnn", resource)
if not data:
data = _load_model_from_github(resource)

# write to cache
with open(cache_path, 'wb') as f:
with open(cache_path, "wb") as f:
f.write(data)

return str(cache_path)
Expand All @@ -305,4 +334,6 @@ def _load_model_from_github(resource):
except Exception:
pass

raise FileNotFoundError(f"Failed to download model from {url}: {type(e).__name__}: {e}")
raise FileNotFoundError(
f"Failed to download model from {url}: {type(e).__name__}: {e}"
)
Loading

0 comments on commit 0808253

Please sign in to comment.