def __init__(self):
super(IRITStartSeg, self).__init__()
self._buffer = BufferTable()
# self.energy = []
self.max_energy = 0.002 * 2
self.min_overlap = 20
self.threshold = 0.12
def __init__(self):
super(IRITStartSeg, self).__init__()
self._buffer = BufferTable()
# self.energy = []
self.max_energy = 0.002*2
self.min_overlap = 20
self.threshold = 0.12
class IRITStartSeg(Analyzer):
'''
Segmentation of recording sessions into 'start' and 'session' segments
Properties:
'''
implements(IAnalyzer)
@interfacedoc
def __init__(self):
super(IRITStartSeg, self).__init__()
self._buffer = BufferTable()
# self.energy = []
self.max_energy = 0.002*2
self.min_overlap = 20
self.threshold = 0.12
@interfacedoc
def setup(self, channels=None, samplerate=None,
blocksize=None, totalframes=None):
super(IRITStartSeg, self).setup(channels,
samplerate,
blocksize,
totalframes)
self.input_blocksize = int(0.02 * samplerate)
self.input_stepsize = int(0.008 * samplerate)
sr = float(samplerate)
lowFreq = 100.0
highFreq = sr / 5
f1 = lowFreq / sr
f2 = highFreq / sr
numtaps = 10
self.filtre = firwin(numtaps=numtaps, cutoff=[f1, f2], pass_zero=False)
self.filtre_z = lfiltic(b=self.filtre, a=1, y=0) # Initial conditions
@staticmethod
@interfacedoc
def id():
return "irit_startseg"
@staticmethod
@interfacedoc
def name():
return "IRIT Start/Session segmentation"
@staticmethod
@interfacedoc
def unit():
return ""
def __str__(self):
return "Labeled Start/session segments"
@frames_adapter
def process(self, frames, eod=False):
'''
'''
#self.energy += [numpy.sqrt(numpy.mean(lfilter(self.filtre,
# 1.0,
# frames.T[0]) ** 2))]
# Compute energy
env, self.filtre_z = lfilter(b=self.filtre, a=1.0, axis=0,
x=frames[:, 0],
zi=self.filtre_z)
self._buffer.append('energy', numpy.sqrt(numpy.mean(env ** 2)))
return frames, eod
def post_process(self):
'''
'''
# Normalize energy
self.energy = self._buffer['energy'][:]
# BAD PATCH !!!
self.energy[-1] = 0
if self.energy.max():
self.energy = self.energy / self.energy.max()
silences = [1 if e < self.max_energy else 0 for e in self.energy]
step = float(self.input_stepsize) / float(self.samplerate())
path = os.path.split(__file__)[0]
models_dir = os.path.join(path, 'trained_models')
prototype1_file = os.path.join(models_dir,
'irit_noise_startSilences_proto1.dat')
prototype2_file = os.path.join(models_dir,
'irit_noise_startSilences_proto2.dat')
prototype = numpy.load(prototype1_file)
prototype2 = numpy.load(prototype2_file)
# Lissage pour éliminer les petits segments dans un sens ou l'autre
struct = [1] * len(prototype)
silences = binary_closing(silences, struct)
silences = binary_opening(silences, struct)
seg = [0, -1, silences[0]]
silencesList = []
for i, v in enumerate(silences):
if not (v == seg[2]):
seg[1] = i
silencesList.append(tuple(seg))
seg = [i, -1, v]
seg[1] = i
silencesList.append(tuple(seg))
segments = []
start = 0.0
for s in silencesList:
if s[2] == 1:
shape = numpy.array(self.energy[s[0]:s[1]])
d1, _ = computeDist2(prototype, shape)
d2, _ = computeDist2(prototype2, shape)
dist = min([d1, d2])
if dist < self.threshold:
s = map(float, s)
segments += [(start, s[0]*step-start, 1) , (s[0]*step, (s[1]-s[0])*step, 0)]
start = s[1]*step
segments += [(start, len(self.energy)*step-start, 1)]
label = {0: 'Start', 1: 'Session'}
segs = self.new_result(data_mode='label', time_mode='segment')
segs.id_metadata.id += '.' + 'segments'
segs.id_metadata.name += ' ' + 'Segments'
segs.data_object.label_metadata.label = label
segs.data_object.time, segs.data_object.duration, segs.data_object.label= zip(*segments)
self.add_result(segs)
def release(self):
self._buffer.close()
def __init__(self,
input_blocksize=2048,
input_stepsize=None,
fft_size=None):
super(SpectrogramBuffer, self).__init__()
self.values = BufferTable()
class SpectrogramBuffer(Spectrogram):
"""
Spectrogram image builder with an extensible buffer based on tables
Parameters
----------
input_blocksize : int, optional
Blocksize of the input signal, default to 2048
input_stepsize : str, optional
The second parameter, default to half blocksize.
fft_size : int, optional
The size of the fft, default to blocksize.
Examples
--------
>>> import timeside
>>> from timeside.core import get_processor
>>> from timeside.core.tools.test_samples import samples
>>> audio_source = samples['sweep.wav']
>>> decoder = get_processor('file_decoder')(uri=audio_source)
>>> spectrogram = get_processor('spectrogram_analyzer_buffer')(input_blocksize=2048, input_stepsize=1024)
>>> pipe = (decoder | spectrogram)
>>> pipe.run()
>>> spectrogram.results.keys()
['spectrogram_analyzer_buffer']
>>> result = spectrogram.results['spectrogram_analyzer_buffer']
>>> result.data.shape
(344, 1025)
.. plot::
import timeside
from timeside.core import get_processor
from timeside.core.tools.test_samples import samples
audio_source = samples['sweep.wav']
decoder = get_processor('file_decoder')(uri=audio_source)
spectrogram = get_processor('spectrogram_analyzer_buffer')(input_blocksize=2048,
input_stepsize=1024)
pipe = (decoder | spectrogram)
pipe.run()
res = spectrogram.results['spectrogram_analyzer_buffer']
res.render()
"""
implements(IAnalyzer)
def __init__(self,
input_blocksize=2048,
input_stepsize=None,
fft_size=None):
super(SpectrogramBuffer, self).__init__()
self.values = BufferTable()
@staticmethod
@interfacedoc
def id():
return "spectrogram_analyzer_buffer"
@staticmethod
@interfacedoc
def name():
return "Spectrogram Analyzer with extensible buffer"
@downmix_to_mono
@frames_adapter
def process(self, frames, eod=False):
stft = np.fft.rfft(frames, self.fft_size)
self.values.append('stft', stft)
return frames, eod
def post_process(self):
spectrogram = self.new_result(data_mode='value', time_mode='framewise')
spectrogram.parameters = {'fft_size': self.fft_size}
spectrogram.data_object.value = np.abs(self.values['stft'])
nb_freq = spectrogram.data_object.value.shape[1]
spectrogram.data_object.y_value = (np.arange(0, nb_freq) *
self.samplerate() / self.fft_size)
self.add_result(spectrogram)
def release(self):
self.values.close()
def __init__(self, input_blocksize=2048, input_stepsize=None,
fft_size=None):
super(SpectrogramBuffer, self).__init__()
self.values = BufferTable()
class SpectrogramBuffer(Spectrogram):
"""
Spectrogram image builder with an extensible buffer based on tables
Parameters
----------
input_blocksize : int, optional
Blocksize of the input signal, default to 2048
input_stepsize : str, optional
The second parameter, default to half blocksize.
fft_size : int, optional
The size of the fft, default to blocksize.
Examples
--------
>>> import timeside
>>> from timeside.core import get_processor
>>> from timeside.core.tools.test_samples import samples
>>> audio_source = samples['sweep.wav']
>>> decoder = get_processor('file_decoder')(uri=audio_source)
>>> spectrogram = get_processor('spectrogram_analyzer_buffer')(input_blocksize=2048, input_stepsize=1024)
>>> pipe = (decoder | spectrogram)
>>> pipe.run()
>>> spectrogram.results.keys()
['spectrogram_analyzer_buffer']
>>> result = spectrogram.results['spectrogram_analyzer_buffer']
>>> result.data.shape
(344, 1025)
.. plot::
import timeside
from timeside.core import get_processor
from timeside.core.tools.test_samples import samples
audio_source = samples['sweep.wav']
decoder = get_processor('file_decoder')(uri=audio_source)
spectrogram = get_processor('spectrogram_analyzer_buffer')(input_blocksize=2048,
input_stepsize=1024)
pipe = (decoder | spectrogram)
pipe.run()
res = spectrogram.results['spectrogram_analyzer_buffer']
res.render()
"""
implements(IAnalyzer)
def __init__(self, input_blocksize=2048, input_stepsize=None,
fft_size=None):
super(SpectrogramBuffer, self).__init__()
self.values = BufferTable()
@staticmethod
@interfacedoc
def id():
return "spectrogram_analyzer_buffer"
@staticmethod
@interfacedoc
def name():
return "Spectrogram Analyzer with extensible buffer"
@downmix_to_mono
@frames_adapter
def process(self, frames, eod=False):
stft = np.fft.rfft(frames, self.fft_size)
self.values.append('stft', stft)
return frames, eod
def post_process(self):
spectrogram = self.new_result(data_mode='value', time_mode='framewise')
spectrogram.parameters = {'fft_size': self.fft_size}
spectrogram.data_object.value = np.abs(self.values['stft'])
nb_freq = spectrogram.data_object.value.shape[1]
spectrogram.data_object.y_value = (np.arange(0, nb_freq) *
self.samplerate() / self.fft_size)
self.add_result(spectrogram)
def release(self):
self.values.close()
class IRITStartSeg(Analyzer):
'''
Segmentation of recording sessions into 'start' and 'session' segments
Properties:
'''
implements(IAnalyzer)
@interfacedoc
def __init__(self):
super(IRITStartSeg, self).__init__()
self._buffer = BufferTable()
# self.energy = []
self.max_energy = 0.002 * 2
self.min_overlap = 20
self.threshold = 0.12
@interfacedoc
def setup(self,
channels=None,
samplerate=None,
blocksize=None,
totalframes=None):
super(IRITStartSeg, self).setup(channels, samplerate, blocksize,
totalframes)
self.input_blocksize = int(0.02 * samplerate)
self.input_stepsize = int(0.008 * samplerate)
sr = float(samplerate)
lowFreq = 100.0
highFreq = sr / 5
f1 = lowFreq / sr
f2 = highFreq / sr
numtaps = 10
self.filtre = firwin(numtaps=numtaps, cutoff=[f1, f2], pass_zero=False)
self.filtre_z = lfiltic(b=self.filtre, a=1, y=0) # Initial conditions
@staticmethod
@interfacedoc
def id():
return "irit_startseg"
@staticmethod
@interfacedoc
def name():
return "IRIT Start/Session segmentation"
@staticmethod
@interfacedoc
def unit():
return ""
def __str__(self):
return "Labeled Start/session segments"
@frames_adapter
def process(self, frames, eod=False):
'''
'''
#self.energy += [numpy.sqrt(numpy.mean(lfilter(self.filtre,
# 1.0,
# frames.T[0]) ** 2))]
# Compute energy
env, self.filtre_z = lfilter(b=self.filtre,
a=1.0,
axis=0,
x=frames[:, 0],
zi=self.filtre_z)
self._buffer.append('energy', numpy.sqrt(numpy.mean(env**2)))
return frames, eod
def post_process(self):
'''
'''
# Normalize energy
self.energy = self._buffer['energy'][:]
# BAD PATCH !!!
self.energy[-1] = 0
if self.energy.max():
self.energy = self.energy / self.energy.max()
silences = [1 if e < self.max_energy else 0 for e in self.energy]
step = float(self.input_stepsize) / float(self.samplerate())
path = os.path.split(__file__)[0]
models_dir = os.path.join(path, 'trained_models')
prototype1_file = os.path.join(models_dir,
'irit_noise_startSilences_proto1.dat')
prototype2_file = os.path.join(models_dir,
'irit_noise_startSilences_proto2.dat')
prototype = numpy.load(prototype1_file)
prototype2 = numpy.load(prototype2_file)
# Lissage pour éliminer les petits segments dans un sens ou l'autre
struct = [1] * len(prototype)
silences = binary_closing(silences, struct)
silences = binary_opening(silences, struct)
seg = [0, -1, silences[0]]
silencesList = []
for i, v in enumerate(silences):
if not (v == seg[2]):
seg[1] = i
silencesList.append(tuple(seg))
seg = [i, -1, v]
seg[1] = i
silencesList.append(tuple(seg))
segments = []
start = 0.0
for s in silencesList:
if s[2] == 1:
shape = numpy.array(self.energy[s[0]:s[1]])
d1, _ = computeDist2(prototype, shape)
d2, _ = computeDist2(prototype2, shape)
dist = min([d1, d2])
if dist < self.threshold:
s = map(float, s)
segments += [(start, s[0] * step - start, 1),
(s[0] * step, (s[1] - s[0]) * step, 0)]
start = s[1] * step
segments += [(start, len(self.energy) * step - start, 1)]
label = {0: 'Start', 1: 'Session'}
segs = self.new_result(data_mode='label', time_mode='segment')
segs.id_metadata.id += '.' + 'segments'
segs.id_metadata.name += ' ' + 'Segments'
segs.data_object.label_metadata.label = label
segs.data_object.time, segs.data_object.duration, segs.data_object.label = zip(
*segments)
self.add_result(segs)
def release(self):
self._buffer.close()
