From 2137c503204fa7108492ea426fab804d411c3dcc Mon Sep 17 00:00:00 2001
From: MasterPhooey <106418429+MasterPhooey@users.noreply.github.com>
Date: Sun, 5 Jan 2025 14:22:44 -0600
Subject: [PATCH] Update advanced_training_notebook.ipynb

---
 advanced_training_notebook.ipynb | 15 +++++----------
 1 file changed, 5 insertions(+), 10 deletions(-)

diff --git a/advanced_training_notebook.ipynb b/advanced_training_notebook.ipynb
index fb479e4..fbf49c3 100644
--- a/advanced_training_notebook.ipynb
+++ b/advanced_training_notebook.ipynb
@@ -177,9 +177,8 @@
     "\n",
     "# Directory setup\n",
     "audioset_dir = \"./audioset\"\n",
-    "os.makedirs(audioset_dir, exist_ok=True)\n",
-    "\n",
     "output_dir = \"./audioset_16k\"\n",
+    "os.makedirs(audioset_dir, exist_ok=True)\n",
     "os.makedirs(output_dir, exist_ok=True)\n",
     "\n",
     "# Full-scale dataset download links\n",
@@ -215,22 +214,17 @@
     "            name = Path(row[\"audio\"][\"path\"]).stem + \".wav\"\n",
     "            output_path = Path(output_dir) / name\n",
     "\n",
-    "            # Read and convert audio\n",
+    "            # Check if audio data is valid before writing\n",
     "            if row[\"audio\"][\"array\"] is None or len(row[\"audio\"][\"array\"]) == 0:\n",
     "                raise ValueError(f\"Empty or invalid audio data in file: {row['audio']['path']}\")\n",
     "\n",
-    "            # Ensure array data is valid before writing\n",
-    "            array_data = row[\"audio\"][\"array\"]\n",
-    "            if not isinstance(array_data, np.ndarray) or len(array_data.shape) == 0:\n",
-    "                raise ValueError(f\"Invalid array data in file: {row['audio']['path']}\")\n",
-    "\n",
     "            scipy.io.wavfile.write(\n",
     "                output_path,\n",
     "                16000,\n",
-    "                (array_data * 32767).astype(np.int16),\n",
+    "                (row[\"audio\"][\"array\"] * 32767).astype(np.int16),\n",
     "            )\n",
     "        except (sf.LibsndfileError, ValueError, Exception) as e:\n",
-    "            # Log and skip problematic files\n",
+    "            # Log the error and skip the file\n",
     "            print(f\"Error converting {row['audio']['path']}: {e}\")\n",
     "            corrupted_files.append(row[\"audio\"][\"path\"])\n",
     "\n",
@@ -245,6 +239,7 @@
     "\n",
     "print(\"Audioset processing complete!\")\n",
     "\n",
+    "\n",
     "# -----------------------------\n",
     "# Download and Process FMA\n",
     "# -----------------------------\n",