improving pitch extraction and separation visualisation

webbook/melodic_analysis/pitch-extraction.ipynb

@@ -280,13 +280,12 @@
     "saraga_tracks = saraga_carnatic.load_tracks()\n",
     "example = saraga_tracks[\"109_Sri_Raghuvara_Sugunaalaya\"]\n",
     "\n",
-    "# Loading the audio\n",
-    "y, sr = librosa.load(example.audio_path)\n",
-    "y = y[:3*sr]  # Getting just the first 3 seconds\n",
-    "\n",
     "# Predict!\n",
-    "ftanet_pitch_track = ftanet_carnatic.predict(y)\n",
-    "melodia_pitch_track = melodia.extract(y)"
+    "melodia_pitch_track = melodia.extract(example.audio_path)\n",
+    "ftanet_pitch_track = ftanet_carnatic.predict(\n",
+    "    example.audio_path,\n",
+    "    out_step=melodia_pitch_track[1, 0],  # Interpolating to same size\n",
+    ")\n"
    ]
   },
   {
@@ -311,6 +310,7 @@
     "D = librosa.amplitude_to_db(np.abs(librosa.stft(y)), ref=np.max)\n",
     "img = librosa.display.specshow(D, y_axis='linear', x_axis='time', sr=sr, ax=ax);\n",
     "ax.set_ylim(0, 2000)\n",
+    "ax.set_xlim(0.6, 3.5)\n",
     "plt.plot(\n",
     "    melodia_pitch_track[:, 0], melodia_pitch_track[:, 1],\n",
     "    color=\"white\", label=\"Melodia\",\n",

webbook/separation/singing-voice-extraction.ipynb

@@ -107,14 +107,20 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "metadata": {},
+   "metadata": {
+    "tags": [
+     "remove-output"
+    ]
+   },
    "outputs": [],
    "source": [
-    "### Getting 30 seconds and separating\n",
-    "input_mixture = input_mixture[:, :44100*30]\n",
+    "### Getting 20 seconds and separating\n",
+    "input_mixture = input_mixture[:, :44100*20]\n",
     "separation = separation_model.separate(\n",
     "    input_data=input_mixture,\n",
-    "    input_sr=sr\n",
+    "    input_sr=sr,\n",
+    "    cluters=6,\n",
+    "    scheduler=5,\n",
     ")"
    ]
   },