Update nbs

nbs/PytorchAudioInference.ipynb

@@ -87,9 +87,9 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "base_folder = Path('../input/kaggle-pog-series-s01e02')\n",
+    "base_folder = Path(\"../input/kaggle-pog-series-s01e02\")\n",
     "\n",
-    "items = get_files(base_folder, extensions='.ogg')"
+    "items = get_files(base_folder, extensions=\".ogg\")"
    ]
   },
   {
@@ -110,7 +110,7 @@
     }
    ],
    "source": [
-    "items\n"
+    "items"
    ]
   },
   {
@@ -126,24 +126,23 @@
     "\n",
     "def create_spectrogram(filename):\n",
     "    audio, sr = torchaudio.load(filename)\n",
-    "    specgram = torchaudio.transforms.MelSpectrogram(sample_rate=sr, \n",
-    "                                                    n_fft=N_FFT, \n",
-    "                                                    win_length=N_FFT, \n",
-    "                                                    hop_length=HOP_LEN\n",
-    "                                                    ,\n",
-    "                                                    center=True,\n",
-    "                                                    pad_mode=\"reflect\",\n",
-    "                                                    power=2.0,\n",
-    "                                                    norm='slaney',\n",
-    "                                                    onesided=True,\n",
-    "                                                    n_mels=224,\n",
-    "                                                    mel_scale=\"htk\"\n",
-    "                                                   )(audio).mean(axis=0)\n",
+    "    specgram = torchaudio.transforms.MelSpectrogram(\n",
+    "        sample_rate=sr,\n",
+    "        n_fft=N_FFT,\n",
+    "        win_length=N_FFT,\n",
+    "        hop_length=HOP_LEN,\n",
+    "        center=True,\n",
+    "        pad_mode=\"reflect\",\n",
+    "        power=2.0,\n",
+    "        norm=\"slaney\",\n",
+    "        onesided=True,\n",
+    "        n_mels=224,\n",
+    "        mel_scale=\"htk\",\n",
+    "    )(audio).mean(axis=0)\n",
     "    specgram = torchaudio.transforms.AmplitudeToDB()(specgram)\n",
     "    specgram = specgram - specgram.min()\n",
-    "    specgram = specgram/specgram.max()\n",
-    "    \n",
-    "    \n",
+    "    specgram = specgram / specgram.max()\n",
+    "\n",
     "    return specgram"
    ]
   },

requirements.txt

@@ -1,3 +1,4 @@
-fastai
-huggingface_hub
+fastai==2.6.0
+gradio==2.9.4
 torchaudio
+torchvision