From 83d8bae37c0dfe73e51ef38128a055f9aa8fa9c8 Mon Sep 17 00:00:00 2001
From: dscripka <david.scripka@gmail.com>
Date: Tue, 5 Sep 2023 07:58:37 -0400
Subject: [PATCH] Convert clips to 16-bit PCM before saving [skip ci]

---
 notebooks/automatic_model_training.ipynb | 25 ++++++++++++++----------
 1 file changed, 15 insertions(+), 10 deletions(-)

diff --git a/notebooks/automatic_model_training.ipynb b/notebooks/automatic_model_training.ipynb
index eb576ca..bf26a24 100644
--- a/notebooks/automatic_model_training.ipynb
+++ b/notebooks/automatic_model_training.ipynb
@@ -68,12 +68,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 91,
    "id": "8f30acfd",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2023-09-04T13:42:01.183840Z",
-     "start_time": "2023-09-04T13:41:59.752153Z"
+     "end_time": "2023-09-05T11:52:51.331099Z",
+     "start_time": "2023-09-05T11:52:50.994098Z"
     }
    },
    "outputs": [],
@@ -127,8 +127,8 @@
    "id": "58e4811e",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2023-09-04T01:07:17.746749Z",
-     "start_time": "2023-09-04T01:07:17.740846Z"
+     "end_time": "2023-09-05T11:53:57.310392Z",
+     "start_time": "2023-09-05T11:53:54.490518Z"
     }
    },
    "outputs": [],
@@ -141,9 +141,10 @@
     "    os.mkdir(output_dir)\n",
     "rir_dataset = datasets.load_dataset(\"davidscripka/MIT_environmental_impulse_responses\", split=\"train\", streaming=True)\n",
     "\n",
+    "# Save clips to 16-bit PCM wav files\n",
     "for row in tqdm(rir_dataset):\n",
     "    name = row['audio']['path'].split('/')[-1]\n",
-    "    scipy.io.wavfile.write(os.path.join(output_dir, name), 16000, row['audio']['array'])\n"
+    "    scipy.io.wavfile.write(os.path.join(output_dir, name), 16000, (row['audio']['array']*32767).astype(np.int16))\n"
    ]
   },
   {
@@ -157,8 +158,10 @@
     "\n",
     "# Audioset Dataset (https://research.google.com/audioset/dataset/index.html)\n",
     "# Download one part of the audioset .tar files, extract, and convert to 16khz\n",
-    "# For full-scale training, it's reccomended to download the entire dataset, and\n",
-    "# even combine with other background noise datasets (e.g., FSD50k, Freesound, etc.)\n",
+    "# For full-scale training, it's recommended to download the entire dataset from \n",
+    "# https://huggingface.co/datasets/agkphysics/AudioSet, and\n",
+    "# even potentially combine it with other background noise datasets (e.g., FSD50k, Freesound, etc.)\n",
+    "\n",
     "fname = \"bal_train09.tar\"\n",
     "out_dir = f\"audioset/{fname}\"\n",
     "link = \"https://huggingface.co/datasets/agkphysics/AudioSet/resolve/main/\" + fname\n",
@@ -172,11 +175,12 @@
     "if not os.path.exists(output_dir):\n",
     "    os.mkdir(output_dir)\n",
     "\n",
+    "# Save clips to 16-bit PCM wav files\n",
     "audioset_dataset = datasets.Dataset.from_dict({\"audio\": [str(i) for i in Path(\"audioset/audio\").glob(\"**/*.flac\")]})\n",
     "audioset_dataset = audioset_dataset.cast_column(\"audio\", datasets.Audio(sampling_rate=16000))\n",
     "for row in tqdm(audioset_dataset):\n",
     "    name = row['audio']['path'].split('/')[-1].replace(\".flac\", \".wav\")\n",
-    "    scipy.io.wavfile.write(os.path.join(output_dir, name), 16000, row['audio']['array'])\n",
+    "    scipy.io.wavfile.write(os.path.join(output_dir, name), 16000, (row['audio']['array']*32767).astype(np.int16))\n",
     "\n",
     "# Free Music Archive dataset\n",
     "# https://github.com/mdeff/fma\n",
@@ -187,11 +191,12 @@
     "fma_dataset = datasets.load_dataset(\"rudraml/fma\", name=\"small\", split=\"train\", streaming=True)\n",
     "fma_dataset = iter(fma_dataset.cast_column(\"audio\", datasets.Audio(sampling_rate=16000)))\n",
     "\n",
+    "# Save clips to 16-bit PCM wav files\n",
     "n_hours = 1  # use only 1 hour of clips for this example notebook, recommend increasing for full-scale training\n",
     "for i in tqdm(range(n_hours*3600//30)):  # this works because the FMA dataset is all 30 second clips\n",
     "    row = next(fma_dataset)\n",
     "    name = row['audio']['path'].split('/')[-1].replace(\".mp3\", \".wav\")\n",
-    "    scipy.io.wavfile.write(os.path.join(output_dir, name), 16000, row['audio']['array'])\n",
+    "    scipy.io.wavfile.write(os.path.join(output_dir, name), 16000, (row['audio']['array']*32767).astype(np.int16))\n",
     "    i += 1\n",
     "    if i == n_hours*3600//30:\n",
     "        break\n"