Merge pull request #142 from CPJKU/fix_issue_141

Fix custom filterbanks in FilteredSpectrograms
CPJKU · Jun 9, 2016 · 7076096 · 7076096
2 parents 6b34ba9 + d2b668c
commit 7076096
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 5 deletions.
diff --git a/CHANGES.rst b/CHANGES.rst
@@ -22,6 +22,14 @@ Other changes:
 * added classes for onset/note/beat detection with RNNs to `features.*` (#118)
 * converted `madmom.modules` into a Python package (#125)
 
+
+Version 0.13.2 (release date: 2016-06-09)
+-----------------------------------------
+
+This is a bugfix release.
+
+* Fix custom filterbank in FilteredSpectrogram (#142)
+
 Version 0.13.1 (release date: 2016-03-14)
 -----------------------------------------
 

diff --git a/madmom/audio/filters.py b/madmom/audio/filters.py
@@ -1058,13 +1058,13 @@ class MelFilterbank(Filterbank):
 
     def __init__(self, bin_frequencies, num_bands=NUM_BANDS, fmin=FMIN,
                  fmax=FMAX, norm_filters=NORM_FILTERS,
-                 unique_filters=UNIQUE_FILTERS):
+                 unique_filters=UNIQUE_FILTERS, **kwargs):
         # this method is for documentation purposes only
         pass
 
     def __new__(cls, bin_frequencies, num_bands=NUM_BANDS, fmin=FMIN,
                 fmax=FMAX, norm_filters=NORM_FILTERS,
-                unique_filters=UNIQUE_FILTERS):
+                unique_filters=UNIQUE_FILTERS, **kwargs):
         # pylint: disable=arguments-differ
         # get a list of frequencies aligned on the Mel scale
         # request 2 more bands, because these are the edge frequencies
@@ -1112,13 +1112,13 @@ class BarkFilterbank(Filterbank):
 
     def __init__(self, bin_frequencies, num_bands=NUM_BANDS, fmin=FMIN,
                  fmax=FMAX, norm_filters=NORM_FILTERS,
-                 unique_filters=UNIQUE_FILTERS):
+                 unique_filters=UNIQUE_FILTERS, **kwargs):
         # this method is for documentation purposes only
         pass
 
     def __new__(cls, bin_frequencies, num_bands=NUM_BANDS, fmin=FMIN,
                 fmax=FMAX, norm_filters=NORM_FILTERS,
-                unique_filters=UNIQUE_FILTERS):
+                unique_filters=UNIQUE_FILTERS, **kwargs):
         # pylint: disable=arguments-differ
         # get a list of frequencies
         if num_bands == 'normal':

diff --git a/tests/test_audio_spectrogram.py b/tests/test_audio_spectrogram.py
@@ -13,7 +13,8 @@
 from .test_audio_filters import FFT_FREQS_1024, LOG_FILTERBANK_CENTER_FREQS
 
 from madmom.audio.spectrogram import *
-from madmom.audio.filters import Filterbank, LogarithmicFilterbank
+from madmom.audio.filters import (Filterbank, LogarithmicFilterbank,
+                                  MelFilterbank, BarkFilterbank)
 from madmom.audio.stft import ShortTimeFourierTransform
 
 
@@ -224,6 +225,23 @@ def test_values(self):
         self.assertTrue(result.num_bins == 81)
         self.assertTrue(result.num_frames == 281)
 
+    def test_filterbanks(self):
+        # with Mel filterbank
+        result = FilteredSpectrogram(AUDIO_PATH + '/sample.wav',
+                                     filterbank=MelFilterbank, num_bands=40)
+        self.assertTrue(np.allclose(result[0, :6],
+                                    [8.42887115, 17.98174477, 19.50165367,
+                                     6.48194313, 2.96991181, 4.06280804]))
+        self.assertTrue(result.shape == (281, 40))
+        # with Bark filterbank
+        result = FilteredSpectrogram(AUDIO_PATH + '/sample.wav',
+                                     filterbank=BarkFilterbank,
+                                     num_bands='normal')
+        self.assertTrue(np.allclose(result[0, :6],
+                                    [16.42251968, 17.36715126, 2.81979132,
+                                     4.27050114, 3.08699131, 1.50553513]))
+        self.assertTrue(result.shape == (281, 23))
+
     def test_methods(self):
         result = FilteredSpectrogram(AUDIO_PATH + '/sample.wav')
         self.assertIsInstance(result.diff(), SpectrogramDifference)