Merge 61eb016 into abc5ce7

librosa · Jun 28, 2018 · 12decdb · 12decdb
2 parents abc5ce7 + 61eb016
commit 12decdb
Show file tree

Hide file tree

Showing 13 changed files with 102 additions and 78 deletions.
diff --git a/docs/examples/plot_hprss.py b/docs/examples/plot_hprss.py
@@ -30,12 +30,17 @@
 ###############################################
 # Compute the short-time Fourier transform of y
 D = librosa.stft(y)
+D_power = librosa.magphase(D, power=2)[0]
+reference_power = np.max(D_power)
+D_dB = librosa.power_to_db(D_power, ref=rp)
 
 #####################################################
 # Decompose D into harmonic and percussive components
 #
 # :math:`D = D_\text{harmonic} + D_\text{percussive}`
 D_harmonic, D_percussive = librosa.decompose.hpss(D)
+D_harmonic_dB = librosa.power_to_db(librosa.magphase(D_harmonic, power=2)[0], ref=rp)
+D_percussive_dB = librosa.power_to_db(librosa.magphase(D_percussive, power=2)[0], ref=rp)
 
 
 ####################################################################
@@ -47,17 +52,17 @@
 plt.figure(figsize=(12, 8))
 
 plt.subplot(3, 1, 1)
-librosa.display.specshow(librosa.amplitude_to_db(D, ref=rp), y_axis='log')
+librosa.display.specshow(D_dB, y_axis='log')
 plt.colorbar()
 plt.title('Full spectrogram')
 
 plt.subplot(3, 1, 2)
-librosa.display.specshow(librosa.amplitude_to_db(D_harmonic, ref=rp), y_axis='log')
+librosa.display.specshow(D_harmonic_dB, y_axis='log')
 plt.colorbar()
 plt.title('Harmonic spectrogram')
 
 plt.subplot(3, 1, 3)
-librosa.display.specshow(librosa.amplitude_to_db(D_percussive, ref=rp), y_axis='log', x_axis='time')
+librosa.display.specshow(D_percussive_dB, y_axis='log', x_axis='time')
 plt.colorbar()
 plt.title('Percussive spectrogram')
 plt.tight_layout()
@@ -80,9 +85,20 @@
 
 # Let's compute separations for a few different margins and compare the results below
 D_harmonic2, D_percussive2 = librosa.decompose.hpss(D, margin=2)
+D_harmonic2_dB = librosa.power_to_db(librosa.magphase(D_harmonic2, power=2)[0], ref=rp)
+D_percussive2_dB = librosa.power_to_db(librosa.magphase(D_percussive2, power=2)[0], ref=rp)
+
 D_harmonic4, D_percussive4 = librosa.decompose.hpss(D, margin=4)
+D_harmonic4_dB = librosa.power_to_db(librosa.magphase(D_harmonic4, power=2)[0], ref=rp)
+D_percussive4_dB = librosa.power_to_db(librosa.magphase(D_percussive4, power=2)[0], ref=rp)
+
 D_harmonic8, D_percussive8 = librosa.decompose.hpss(D, margin=8)
+D_harmonic8_dB = librosa.power_to_db(librosa.magphase(D_harmonic8, power=2)[0], ref=rp)
+D_percussive8_dB = librosa.power_to_db(librosa.magphase(D_percussive8, power=2)[0], ref=rp)
+
 D_harmonic16, D_percussive16 = librosa.decompose.hpss(D, margin=16)
+D_harmonic16_dB = librosa.power_to_db(librosa.magphase(D_harmonic16, power=2)[0], ref=rp)
+D_percussive16_dB = librosa.power_to_db(librosa.magphase(D_percussive16, power=2)[0], ref=rp)
 
 
 #############################################################################
@@ -91,41 +107,41 @@
 plt.figure(figsize=(10, 10))
 
 plt.subplot(5, 2, 1)
-librosa.display.specshow(librosa.amplitude_to_db(D_harmonic, ref=rp), y_axis='log')
+librosa.display.specshow(D_harmonic_dB, y_axis='log')
 plt.title('Harmonic')
 plt.yticks([])
 plt.ylabel('margin=1')
 
 plt.subplot(5, 2, 2)
-librosa.display.specshow(librosa.amplitude_to_db(D_percussive, ref=rp), y_axis='log')
+librosa.display.specshow(D_percussive_dB, y_axis='log')
 plt.title('Percussive')
 plt.yticks([]), plt.ylabel('')
 
 plt.subplot(5, 2, 3)
-librosa.display.specshow(librosa.amplitude_to_db(D_harmonic2, ref=rp), y_axis='log')
+librosa.display.specshow(D_harmonic2_dB, y_axis='log')
 plt.yticks([])
 plt.ylabel('margin=2')
 
 plt.subplot(5, 2, 4)
-librosa.display.specshow(librosa.amplitude_to_db(D_percussive2, ref=rp), y_axis='log')
+librosa.display.specshow(D_percussive2_dB, y_axis='log')
 plt.yticks([]) ,plt.ylabel('')
 
 plt.subplot(5, 2, 5)
-librosa.display.specshow(librosa.amplitude_to_db(D_harmonic4, ref=rp), y_axis='log')
+librosa.display.specshow(D_harmonic4_dB, y_axis='log')
 plt.yticks([])
 plt.ylabel('margin=4')
 
 plt.subplot(5, 2, 6)
-librosa.display.specshow(librosa.amplitude_to_db(D_percussive4, ref=rp), y_axis='log')
+librosa.display.specshow(D_percussive4_dB, y_axis='log')
 plt.yticks([]), plt.ylabel('')
 
 plt.subplot(5, 2, 7)
-librosa.display.specshow(librosa.amplitude_to_db(D_harmonic8, ref=rp), y_axis='log')
+librosa.display.specshow(D_harmonic8_dB, y_axis='log')
 plt.yticks([])
 plt.ylabel('margin=8')
 
 plt.subplot(5, 2, 8)
-librosa.display.specshow(librosa.amplitude_to_db(D_percussive8, ref=rp), y_axis='log')
+librosa.display.specshow(D_percussive8_dB, y_axis='log')
 plt.yticks([]), plt.ylabel('')
 
 plt.subplot(5, 2, 9)

diff --git a/docs/examples/plot_presets.py b/docs/examples/plot_presets.py
@@ -59,29 +59,25 @@
 ######################################################################
 # Now we can load in a file and do some analysis with the new defaults
 filename = 'audio/Karissa_Hobbs_-_09_-_Lets_Go_Fishin.mp3'
-
 y, sr = librosa.load(filename, duration=5, offset=35)
 
-# Generate a Mel spectrogram:
-
+# Generate a mel spectrogram:
 M = librosa.feature.melspectrogram(y=y)
 
 # Of course, you can still override the new default manually, e.g.:
-
-M_highres = librosa.feature.melspectrogram(y=y, hop_length=512)
-
+M_hop512 = librosa.feature.melspectrogram(y=y, hop_length=512)
 
 # And plot the results
 plt.figure(figsize=(6, 6))
 ax = plt.subplot(3, 1, 1)
 
-librosa.display.specshow(librosa.power_to_db(M, ref=np.max),
+librosa.display.specshow(librosa.amplitude_to_db(M, ref=np.max),
                          y_axis='mel', x_axis='time')
 
 plt.title('44100/1024/4096')
 
 plt.subplot(3, 1, 2, sharex=ax, sharey=ax)
-librosa.display.specshow(librosa.power_to_db(M_highres, ref=np.max),
+librosa.display.specshow(librosa.amplitude_to_db(M_hop512, ref=np.max),
                          hop_length=512,
                          y_axis='mel', x_axis='time')
 plt.title('44100/512/4096')
@@ -91,10 +87,10 @@
 librosa['sr'] = 11025
 
 y2, sr2 = librosa.load(filename, duration=5, offset=35)
-M2 = librosa.feature.melspectrogram(y=y2, sr=sr2)
+M_11k = librosa.feature.melspectrogram(y=y2, sr=sr2)
 
 plt.subplot(3, 1, 3, sharex=ax, sharey=ax)
-librosa.display.specshow(librosa.power_to_db(M2, ref=np.max),
+librosa.display.specshow(librosa.amplitude_to_db(M_11k, ref=np.max),
                          y_axis='mel', x_axis='time')
 
 plt.title('11025/1024/4096')

diff --git a/docs/examples/plot_segmentation.py b/docs/examples/plot_segmentation.py
@@ -43,13 +43,14 @@
 # Next, we'll compute and plot a log-power CQT
 BINS_PER_OCTAVE = 12 * 3
 N_OCTAVES = 7
-C = librosa.amplitude_to_db(librosa.cqt(y=y, sr=sr,
-                                        bins_per_octave=BINS_PER_OCTAVE,
-                                        n_bins=N_OCTAVES * BINS_PER_OCTAVE),
-                            ref=np.max)
+C = librosa.cqt(y=y, sr=sr,
+                bins_per_octave=BINS_PER_OCTAVE,
+                n_bins=N_OCTAVES * BINS_PER_OCTAVE),
+    ref=np.max)
+C_dB = librosa.power_to_db(librosa.magphase(C, power=2)[0])
 
 plt.figure(figsize=(12, 4))
-librosa.display.specshow(C, y_axis='cqt_hz', sr=sr,
+librosa.display.specshow(C_dB, y_axis='cqt_hz', sr=sr,
                          bins_per_octave=BINS_PER_OCTAVE,
                          x_axis='time')
 plt.tight_layout()
@@ -58,7 +59,7 @@
 ##########################################################
 # To reduce dimensionality, we'll beat-synchronous the CQT
 tempo, beats = librosa.beat.beat_track(y=y, sr=sr, trim=False)
-Csync = librosa.util.sync(C, beats, aggregate=np.median)
+Csync_dB = librosa.util.sync(C_dB, beats, aggregate=np.median)
 
 # For plotting purposes, we'll need the timing of the beats
 # we fix_frames to include non-beat frames 0 and C.shape[1] (final frame)
@@ -68,7 +69,7 @@
                                     sr=sr)
 
 plt.figure(figsize=(12, 4))
-librosa.display.specshow(Csync, bins_per_octave=12*3,
+librosa.display.specshow(Csync_dB, bins_per_octave=12*3,
                          y_axis='cqt_hz', x_axis='time',
                          x_coords=beat_times)
 plt.tight_layout()
@@ -79,7 +80,7 @@
 # (Equation 1)
 # width=3 prevents links within the same bar
 # mode='affinity' here implements S_rep (after Eq. 8)
-R = librosa.segment.recurrence_matrix(Csync, width=3, mode='affinity',
+R = librosa.segment.recurrence_matrix(Csync_dB, width=3, mode='affinity',
                                       sym=True)
 
 # Enhance diagonals with a median filter (Equation 2)
@@ -230,7 +231,7 @@
                                 fmin=librosa.note_to_hz('C1'),
                                 bins_per_octave=BINS_PER_OCTAVE)
 
-librosa.display.specshow(C, y_axis='cqt_hz', sr=sr,
+librosa.display.specshow(C_dB, y_axis='cqt_hz', sr=sr,
                          bins_per_octave=BINS_PER_OCTAVE,
                          x_axis='time')
 ax = plt.gca()

diff --git a/docs/examples/plot_superflux.py b/docs/examples/plot_superflux.py
@@ -55,8 +55,12 @@
                                    n_mels=n_mels)
 
 
+# Extract magnitude and convert to dB
+S_dB = librosa.amplitude_to_db(librosa.magphase(S)[0], ref=np.max)
+
+
 plt.figure(figsize=(6, 4))
-librosa.display.specshow(librosa.power_to_db(S, ref=np.max),
+librosa.display.specshow(S_dB,
                          y_axis='mel', x_axis='time', sr=sr,
                          hop_length=hop_length, fmin=fmin, fmax=fmax)
 plt.tight_layout()
@@ -72,7 +76,7 @@
 
 #########################################
 # And similarly with the superflux method
-odf_sf = librosa.onset.onset_strength(S=librosa.power_to_db(S, ref=np.max),
+odf_sf = librosa.onset.onset_strength(S=S_dB,
                                       sr=sr,
                                       hop_length=hop_length,
                                       lag=lag, max_size=max_size)
@@ -100,7 +104,7 @@
                                     hop_length=hop_length)
 
 ax = plt.subplot(2, 1, 2)
-librosa.display.specshow(librosa.power_to_db(S, ref=np.max),
+librosa.display.specshow(S_dB,
                          y_axis='mel', x_axis='time', sr=sr,
                          hop_length=hop_length, fmin=fmin, fmax=fmax)
 plt.xlim([0, 5.0])

diff --git a/examples/LibROSA audio effects and playback.ipynb b/examples/LibROSA audio effects and playback.ipynb
@@ -256,7 +256,8 @@
     "plt.figure(figsize=(12,4))\n",
     "\n",
     "plt.subplot(1,2,1)\n",
-    "librosa.display.specshow(librosa.amplitude_to_db(components, ref=np.max), y_axis='log')\n",
+    "components_db = librosa.amplitude_to_db(np.abs(components), ref=np.max)\n",
+    "librosa.display.specshow(components_db, y_axis='log')\n",
     "plt.xlabel('Component')\n",
     "plt.ylabel('Frequency')\n",
     "plt.title('Components')\n",

diff --git a/examples/LibROSA demo.ipynb b/examples/LibROSA demo.ipynb
@@ -108,11 +108,12 @@
     }
    ],
    "source": [
-    "# Let's make and display a mel-scaled power (energy-squared) spectrogram\n",
+    "# Let's make and display a mel-scaled power (squared energy) spectrogram\n",
     "S = librosa.feature.melspectrogram(y, sr=sr, n_mels=128)\n",
+    "S_power = librosa.magphase(S, power=2)[0]\n",
     "\n",
     "# Convert to log scale (dB). We'll use the peak power (max) as reference.\n",
-    "log_S = librosa.power_to_db(S, ref=np.max)\n",
+    "log_S = librosa.amplitude_to_db(S_power, ref=np.max)\n",
     "\n",
     "# Make a new figure\n",
     "plt.figure(figsize=(12,4))\n",
@@ -167,13 +168,15 @@
    ],
    "source": [
     "# What do the spectrograms look like?\n",
-    "# Let's make and display a mel-scaled power (energy-squared) spectrogram\n",
+    "# Let's make and display a mel-scaled power (squared energy) spectrogram\n",
     "S_harmonic   = librosa.feature.melspectrogram(y_harmonic, sr=sr)\n",
     "S_percussive = librosa.feature.melspectrogram(y_percussive, sr=sr)\n",
+    "Sh_power = librosa.magphase(S_harmonic, power=2)[0]\n",
+    "Sp_power = librosa.magphase(S_percussive, power=2)[0]\n",
     "\n",
     "# Convert to log scale (dB). We'll use the peak power as reference.\n",
-    "log_Sh = librosa.power_to_db(S_harmonic, ref=np.max)\n",
-    "log_Sp = librosa.power_to_db(S_percussive, ref=np.max)\n",
+    "log_Sh = librosa.power_to_db(Sh_power, ref=np.max)\n",
+    "log_Sp = librosa.power_to_db(Sp_power, ref=np.max)\n",
     "\n",
     "# Make a new figure\n",
     "plt.figure(figsize=(12,6))\n",

diff --git a/librosa/core/audio.py b/librosa/core/audio.py
@@ -647,8 +647,9 @@ def clicks(times=None, frames=None, sr=22050, hop_length=512,
     >>> import matplotlib.pyplot as plt
     >>> plt.figure()
     >>> S = librosa.feature.melspectrogram(y=y, sr=sr)
+    >>> S_power = librosa.magphase(S, power=2)[0]
     >>> ax = plt.subplot(2,1,2)
-    >>> librosa.display.specshow(librosa.power_to_db(S, ref=np.max),
+    >>> librosa.display.specshow(librosa.power_to_db(S_power, ref=np.max),
     ...                          x_axis='time', y_axis='mel')
     >>> plt.subplot(2,1,1, sharex=ax)
     >>> librosa.display.waveplot(y_beat_times, sr=sr, label='Beat clicks')

diff --git a/librosa/core/constantq.py b/librosa/core/constantq.py
@@ -119,7 +119,8 @@ def cqt(y, sr=22050, hop_length=512, fmin=None, n_bins=84,
     >>> import matplotlib.pyplot as plt
     >>> y, sr = librosa.load(librosa.util.example_audio_file())
     >>> C = librosa.cqt(y, sr=sr)
-    >>> librosa.display.specshow(librosa.amplitude_to_db(C, ref=np.max),
+    >>> C_dB = librosa.power_to_db(librosa.magphase(C, power=2)[0], ref=np.max)
+    >>> librosa.display.specshow(C_dB,
     ...                          sr=sr, x_axis='time', y_axis='cqt_note')
     >>> plt.colorbar(format='%+2.0f dB')
     >>> plt.title('Constant-Q power spectrum')