class Document(rs):
windowed = ArrayWithUnitsFeature(SlidingWindow,
wscheme=wscheme,
needs=rs.resampled,
store=False)
dct = ArrayWithUnitsFeature(DCTIV, needs=windowed, store=True)
class ShortTimeFourierTransform(BaseModel):
meta = JSONFeature(MetaData, store=True, encoder=AudioMetaDataEncoder)
raw = ByteStreamFeature(ByteStream,
chunksize=chunksize_bytes,
needs=meta,
store=False)
ogg = OggVorbisFeature(OggVorbis, needs=raw, store=True)
pcm = AudioSamplesFeature(AudioStream, needs=raw, store=False)
resampled = AudioSamplesFeature(Resampler,
needs=pcm,
samplerate=resample_to,
store=store_resampled)
windowed = ArrayWithUnitsFeature(SlidingWindow,
needs=resampled,
wscheme=wscheme,
wfunc=OggVorbisWindowingFunc(),
store=store_windowed)
fft = ArrayWithUnitsFeature(FFT,
padding_samples=fft_padding_samples,
needs=windowed,
store=store_fft)
class Sound(rs):
short_windowed = ArrayWithUnitsFeature(SlidingWindow,
wscheme=short_window,
needs=rs.resampled)
long_windowed = ArrayWithUnitsFeature(SlidingWindow,
wscheme=long_window,
needs=short_windowed)
class Document(BaseModel, PersistenceSettings):
source1 = ArrayWithUnitsFeature(MergeTester,
total_frames=200,
store=True)
source2 = ArrayWithUnitsFeature(MergeTester,
total_frames=200,
store=True)
merged = ArrayWithUnitsFeature(Merge,
needs=[source1, source2],
store=True)
class Document(rs):
long_windowed = ArrayWithUnitsFeature(
SlidingWindow,
wscheme=SampleRate(Milliseconds(500), Seconds(1)),
wfunc=OggVorbisWindowingFunc(),
needs=rs.resampled,
store=True)
long_fft = ArrayWithUnitsFeature(FFT,
needs=long_windowed,
store=True)
class Document(rs):
windowed = ArrayWithUnitsFeature(SlidingWindow,
wscheme=wscheme,
wfunc=OggVorbisWindowingFunc(),
needs=rs.resampled,
store=False)
fft = ArrayWithUnitsFeature(FFT, needs=windowed, store=False)
centroid = ArrayWithUnitsFeature(SpectralCentroid,
needs=fft,
store=True)
class Model(STFT, settings):
binary = ArrayWithUnitsFeature(Binarize,
predicate=lambda data: data >= 0,
needs=STFT.fft,
store=False)
sliced = ArrayWithUnitsFeature(Slice,
sl=slice(0, slice_size),
needs=binary,
store=True)
packed = ArrayWithUnitsFeature(pack, needs=sliced, store=True)
class WithOnsets(self.STFT):
transience = ArrayWithUnitsFeature(MeasureOfTransience,
needs=self.STFT.fft,
store=True)
sliding_detection = ArrayWithUnitsFeature(
SlidingWindow,
needs=transience,
wscheme=self.wscheme * Stride(frequency=1, duration=10),
padwith=5,
store=False)
slices = TimeSliceFeature(MovingAveragePeakPicker,
needs=sliding_detection,
aggregate=np.median,
store=True)
class Document(BaseModel, PersistenceSettings):
source1 = ArrayWithUnitsFeature(MergeTester,
total_frames=200,
increments_of=12,
store=True)
source2 = ArrayWithUnitsFeature(MergeTester,
total_frames=200,
increments_of=17,
store=True)
source3 = ArrayWithUnitsFeature(MergeTester,
total_frames=200,
increments_of=32,
store=True)
merged = ArrayWithUnitsFeature(Merge,
needs=(source1, source2, source3),
store=True)
class FrequencyAdaptive(BaseModel):
long_windowed = ArrayWithUnitsFeature(SlidingWindow,
wscheme=long_window_sample_rate,
wfunc=OggVorbisWindowingFunc(),
needs=BaseModel.resampled,
store=False)
long_fft = ArrayWithUnitsFeature(FFT, needs=long_windowed, store=False)
freq_adaptive = FrequencyAdaptiveFeature(
FrequencyAdaptiveTransform,
transform=np.fft.irfft,
scale=scale,
check_scale_overlap_ratio=check_scale_overlap_ratio,
window_func=np.hanning,
needs=long_fft,
store=store_freq_adaptive)
class Document(BaseModel):
long_windowed = ArrayWithUnitsFeature(
SlidingWindow,
wscheme=SampleRate(frequency=Milliseconds(500),
duration=Seconds(1)),
wfunc=windowing_func,
needs=BaseModel.resampled,
store=True)
dct = ArrayWithUnitsFeature(DCT,
scale_always_even=True,
needs=long_windowed,
store=True)
mdct = FrequencyAdaptiveFeature(FrequencyAdaptiveTransform,
transform=scipy.fftpack.idct,
scale=scale,
needs=dct,
store=True)
class WithOnsets(self.STFT, Settings):
onset_prep = ArrayWithUnitsFeature(SlidingWindow,
needs=self.STFT.fft,
wscheme=self.wscheme * (1, 3),
store=False)
complex_domain = ArrayWithUnitsFeature(ComplexDomain,
needs=onset_prep,
store=False)
sliding_detection = ArrayWithUnitsFeature(SlidingWindow,
needs=complex_domain,
wscheme=self.wscheme *
(1, 11),
padwith=5,
store=False)
slices = TimeSliceFeature(MovingAveragePeakPicker,
needs=sliding_detection,
aggregate=np.median,
store=True)
class Onsets(BaseModel):
onset_prep = ArrayWithUnitsFeature(SlidingWindow,
needs=fft_feature,
wscheme=HalfLapped() *
Stride(frequency=1, duration=3),
store=False)
complex_domain = ArrayWithUnitsFeature(ComplexDomain,
needs=onset_prep,
store=False)
sliding_detection = ArrayWithUnitsFeature(
SlidingWindow,
needs=complex_domain,
wscheme=HalfLapped() * Stride(frequency=1, duration=11),
padwith=5,
store=False)
slices = TimeSliceFeature(MovingAveragePeakPicker,
needs=sliding_detection,
aggregate=np.median,
store=True)
class Document(BaseModel):
raw = ByteStreamFeature(ByteStream,
chunksize=2 * 44100 * 30 * 2,
store=True)
ogg = OggVorbisFeature(OggVorbis, needs=raw, store=True)
pcm = AudioSamplesFeature(AudioStream, needs=raw, store=True)
resampled = AudioSamplesFeature(Resampler,
needs=pcm,
samplerate=samplerate,
store=True)
windowed = ArrayWithUnitsFeature(SlidingWindow,
needs=resampled,
wscheme=windowing_scheme,
wfunc=OggVorbisWindowingFunc(),
store=False)
fft = ArrayWithUnitsFeature(FFT, needs=windowed, store=True)
chroma = ArrayWithUnitsFeature(Chroma,
needs=fft,
frequency_band=band,
store=True)
bark = ArrayWithUnitsFeature(BarkBands,
needs=fft,
frequency_band=band,
store=True)
bfcc = ArrayWithUnitsFeature(BFCC, needs=bark, store=True)
bfcc_sliding_window = ArrayWithUnitsFeature(
SlidingWindow,
needs=bfcc,
wscheme=windowing_scheme * Stride(frequency=2, duration=4),
store=True)
bfcc_pooled = ArrayWithUnitsFeature(Max,
needs=bfcc_sliding_window,
axis=1,
store=True)
class AudioGraph(BaseModel):
meta = JSONFeature(MetaData, store=True, encoder=AudioMetaDataEncoder)
raw = ByteStreamFeature(ByteStream,
chunksize=chunksize_bytes,
needs=meta,
store=False)
ogg = OggVorbisFeature(OggVorbis, needs=raw, store=True)
pcm = AudioSamplesFeature(AudioStream, needs=raw, store=False)
resampled = AudioSamplesFeature(Resampler,
needs=pcm,
samplerate=resample_to,
store=False)
windowed = ArrayWithUnitsFeature(SlidingWindow,
needs=resampled,
wscheme=HalfLapped(),
wfunc=OggVorbisWindowingFunc(),
store=False)
dct = ArrayWithUnitsFeature(DCT, needs=windowed, store=True)
fft = ArrayWithUnitsFeature(FFT, needs=windowed, store=store_fft)
bark = ArrayWithUnitsFeature(BarkBands,
needs=fft,
frequency_band=band,
store=True)
centroid = ArrayWithUnitsFeature(SpectralCentroid,
needs=bark,
store=True)
chroma = ArrayWithUnitsFeature(Chroma,
needs=fft,
frequency_band=band,
store=True)
bfcc = ArrayWithUnitsFeature(BFCC, needs=fft, store=True)
class Document(fa):
rasterized = ArrayWithUnitsFeature(
lambda fa: fa.rasterize(64).astype(np.float32),
needs=fa.freq_adaptive,
store=True)
class Sound(rs):
windowed = ArrayWithUnitsFeature(SlidingWindow,
wscheme=wscheme,
wfunc=wfunc,
needs=rs.resampled,
store=store_windowed)
class Document(BaseModel, PersistenceSettings):
source = ArrayWithUnitsFeature(MergeTester, store=True)
merged = ArrayWithUnitsFeature(Merge, needs=[source], store=True)
class Document(BaseModel):
windowed = ArrayWithUnitsFeature(SlidingWindow,
wscheme=sw,
store=True)
class Document(rs):
windowed = ArrayWithUnitsFeature(SlidingWindow,
wscheme=HalfLapped(),
needs=rs.resampled,
store=True)