Upload ai_music_detection_new_large.ipynb
Browse files
ai_music_detection_new_large.ipynb
ADDED
|
@@ -0,0 +1,819 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cells": [
|
| 3 |
+
{
|
| 4 |
+
"cell_type": "code",
|
| 5 |
+
"execution_count": 1,
|
| 6 |
+
"metadata": {},
|
| 7 |
+
"outputs": [
|
| 8 |
+
{
|
| 9 |
+
"name": "stdout",
|
| 10 |
+
"output_type": "stream",
|
| 11 |
+
"text": [
|
| 12 |
+
"Requirement already satisfied: librosa in /opt/conda/lib/python3.10/site-packages (0.10.2.post1)\n",
|
| 13 |
+
"Requirement already satisfied: soundfile in /opt/conda/lib/python3.10/site-packages (0.12.1)\n",
|
| 14 |
+
"Requirement already satisfied: torchaudio in /opt/conda/lib/python3.10/site-packages (2.2.0)\n",
|
| 15 |
+
"Requirement already satisfied: audiomentations in /opt/conda/lib/python3.10/site-packages (0.37.0)\n",
|
| 16 |
+
"Requirement already satisfied: evaluate in /opt/conda/lib/python3.10/site-packages (0.4.3)\n",
|
| 17 |
+
"Requirement already satisfied: ipywidgets in /opt/conda/lib/python3.10/site-packages (8.1.5)\n",
|
| 18 |
+
"Requirement already satisfied: matplotlib in /opt/conda/lib/python3.10/site-packages (3.9.3)\n",
|
| 19 |
+
"Requirement already satisfied: tensorboard in /opt/conda/lib/python3.10/site-packages (2.18.0)\n",
|
| 20 |
+
"Requirement already satisfied: datasets[audio] in /opt/conda/lib/python3.10/site-packages (3.1.0)\n",
|
| 21 |
+
"Requirement already satisfied: transformers[torch] in /opt/conda/lib/python3.10/site-packages (4.47.0)\n",
|
| 22 |
+
"Requirement already satisfied: audioread>=2.1.9 in /opt/conda/lib/python3.10/site-packages (from librosa) (3.0.1)\n",
|
| 23 |
+
"Requirement already satisfied: numpy!=1.22.0,!=1.22.1,!=1.22.2,>=1.20.3 in /opt/conda/lib/python3.10/site-packages (from librosa) (1.26.3)\n",
|
| 24 |
+
"Requirement already satisfied: scipy>=1.2.0 in /opt/conda/lib/python3.10/site-packages (from librosa) (1.12.0)\n",
|
| 25 |
+
"Requirement already satisfied: scikit-learn>=0.20.0 in /opt/conda/lib/python3.10/site-packages (from librosa) (1.5.2)\n",
|
| 26 |
+
"Requirement already satisfied: joblib>=0.14 in /opt/conda/lib/python3.10/site-packages (from librosa) (1.4.2)\n",
|
| 27 |
+
"Requirement already satisfied: decorator>=4.3.0 in /opt/conda/lib/python3.10/site-packages (from librosa) (5.1.1)\n",
|
| 28 |
+
"Requirement already satisfied: numba>=0.51.0 in /opt/conda/lib/python3.10/site-packages (from librosa) (0.60.0)\n",
|
| 29 |
+
"Requirement already satisfied: pooch>=1.1 in /opt/conda/lib/python3.10/site-packages (from librosa) (1.8.2)\n",
|
| 30 |
+
"Requirement already satisfied: soxr>=0.3.2 in /opt/conda/lib/python3.10/site-packages (from librosa) (0.5.0.post1)\n",
|
| 31 |
+
"Requirement already satisfied: typing-extensions>=4.1.1 in /opt/conda/lib/python3.10/site-packages (from librosa) (4.9.0)\n",
|
| 32 |
+
"Requirement already satisfied: lazy-loader>=0.1 in /opt/conda/lib/python3.10/site-packages (from librosa) (0.4)\n",
|
| 33 |
+
"Requirement already satisfied: msgpack>=1.0 in /opt/conda/lib/python3.10/site-packages (from librosa) (1.1.0)\n",
|
| 34 |
+
"Requirement already satisfied: cffi>=1.0 in /opt/conda/lib/python3.10/site-packages (from soundfile) (1.16.0)\n",
|
| 35 |
+
"Requirement already satisfied: torch in /opt/conda/lib/python3.10/site-packages (from torchaudio) (2.2.0)\n",
|
| 36 |
+
"Requirement already satisfied: filelock in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (3.13.1)\n",
|
| 37 |
+
"Requirement already satisfied: pyarrow>=15.0.0 in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (18.1.0)\n",
|
| 38 |
+
"Requirement already satisfied: dill<0.3.9,>=0.3.0 in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (0.3.8)\n",
|
| 39 |
+
"Requirement already satisfied: pandas in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (2.2.3)\n",
|
| 40 |
+
"Requirement already satisfied: requests>=2.32.2 in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (2.32.3)\n",
|
| 41 |
+
"Requirement already satisfied: tqdm>=4.66.3 in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (4.67.1)\n",
|
| 42 |
+
"Requirement already satisfied: xxhash in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (3.5.0)\n",
|
| 43 |
+
"Requirement already satisfied: multiprocess<0.70.17 in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (0.70.16)\n",
|
| 44 |
+
"Requirement already satisfied: fsspec<=2024.9.0,>=2023.1.0 in /opt/conda/lib/python3.10/site-packages (from fsspec[http]<=2024.9.0,>=2023.1.0->datasets[audio]) (2023.12.2)\n",
|
| 45 |
+
"Requirement already satisfied: aiohttp in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (3.11.10)\n",
|
| 46 |
+
"Requirement already satisfied: huggingface-hub>=0.23.0 in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (0.26.3)\n",
|
| 47 |
+
"Requirement already satisfied: packaging in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (23.1)\n",
|
| 48 |
+
"Requirement already satisfied: pyyaml>=5.1 in /opt/conda/lib/python3.10/site-packages (from datasets[audio]) (6.0.1)\n",
|
| 49 |
+
"Requirement already satisfied: regex!=2019.12.17 in /opt/conda/lib/python3.10/site-packages (from transformers[torch]) (2024.11.6)\n",
|
| 50 |
+
"Requirement already satisfied: tokenizers<0.22,>=0.21 in /opt/conda/lib/python3.10/site-packages (from transformers[torch]) (0.21.0)\n",
|
| 51 |
+
"Requirement already satisfied: safetensors>=0.4.1 in /opt/conda/lib/python3.10/site-packages (from transformers[torch]) (0.4.5)\n",
|
| 52 |
+
"Requirement already satisfied: accelerate>=0.26.0 in /opt/conda/lib/python3.10/site-packages (from transformers[torch]) (1.1.1)\n",
|
| 53 |
+
"Requirement already satisfied: numpy-minmax<1,>=0.3.0 in /opt/conda/lib/python3.10/site-packages (from audiomentations) (0.3.1)\n",
|
| 54 |
+
"Requirement already satisfied: numpy-rms<1,>=0.4.2 in /opt/conda/lib/python3.10/site-packages (from audiomentations) (0.4.2)\n",
|
| 55 |
+
"Requirement already satisfied: comm>=0.1.3 in /opt/conda/lib/python3.10/site-packages (from ipywidgets) (0.2.2)\n",
|
| 56 |
+
"Requirement already satisfied: ipython>=6.1.0 in /opt/conda/lib/python3.10/site-packages (from ipywidgets) (8.20.0)\n",
|
| 57 |
+
"Requirement already satisfied: traitlets>=4.3.1 in /opt/conda/lib/python3.10/site-packages (from ipywidgets) (5.7.1)\n",
|
| 58 |
+
"Requirement already satisfied: widgetsnbextension~=4.0.12 in /opt/conda/lib/python3.10/site-packages (from ipywidgets) (4.0.13)\n",
|
| 59 |
+
"Requirement already satisfied: jupyterlab-widgets~=3.0.12 in /opt/conda/lib/python3.10/site-packages (from ipywidgets) (3.0.13)\n",
|
| 60 |
+
"Requirement already satisfied: contourpy>=1.0.1 in /opt/conda/lib/python3.10/site-packages (from matplotlib) (1.3.1)\n",
|
| 61 |
+
"Requirement already satisfied: cycler>=0.10 in /opt/conda/lib/python3.10/site-packages (from matplotlib) (0.12.1)\n",
|
| 62 |
+
"Requirement already satisfied: fonttools>=4.22.0 in /opt/conda/lib/python3.10/site-packages (from matplotlib) (4.55.2)\n",
|
| 63 |
+
"Requirement already satisfied: kiwisolver>=1.3.1 in /opt/conda/lib/python3.10/site-packages (from matplotlib) (1.4.7)\n",
|
| 64 |
+
"Requirement already satisfied: pillow>=8 in /opt/conda/lib/python3.10/site-packages (from matplotlib) (10.0.1)\n",
|
| 65 |
+
"Requirement already satisfied: pyparsing>=2.3.1 in /opt/conda/lib/python3.10/site-packages (from matplotlib) (3.2.0)\n",
|
| 66 |
+
"Requirement already satisfied: python-dateutil>=2.7 in /opt/conda/lib/python3.10/site-packages (from matplotlib) (2.9.0.post0)\n",
|
| 67 |
+
"Requirement already satisfied: absl-py>=0.4 in /opt/conda/lib/python3.10/site-packages (from tensorboard) (2.1.0)\n",
|
| 68 |
+
"Requirement already satisfied: grpcio>=1.48.2 in /opt/conda/lib/python3.10/site-packages (from tensorboard) (1.68.1)\n",
|
| 69 |
+
"Requirement already satisfied: markdown>=2.6.8 in /opt/conda/lib/python3.10/site-packages (from tensorboard) (3.7)\n",
|
| 70 |
+
"Requirement already satisfied: protobuf!=4.24.0,>=3.19.6 in /opt/conda/lib/python3.10/site-packages (from tensorboard) (5.29.1)\n",
|
| 71 |
+
"Requirement already satisfied: setuptools>=41.0.0 in /opt/conda/lib/python3.10/site-packages (from tensorboard) (68.2.2)\n",
|
| 72 |
+
"Requirement already satisfied: six>1.9 in /opt/conda/lib/python3.10/site-packages (from tensorboard) (1.16.0)\n",
|
| 73 |
+
"Requirement already satisfied: tensorboard-data-server<0.8.0,>=0.7.0 in /opt/conda/lib/python3.10/site-packages (from tensorboard) (0.7.2)\n",
|
| 74 |
+
"Requirement already satisfied: werkzeug>=1.0.1 in /opt/conda/lib/python3.10/site-packages (from tensorboard) (3.1.3)\n",
|
| 75 |
+
"Requirement already satisfied: psutil in /opt/conda/lib/python3.10/site-packages (from accelerate>=0.26.0->transformers[torch]) (5.9.0)\n",
|
| 76 |
+
"Requirement already satisfied: pycparser in /opt/conda/lib/python3.10/site-packages (from cffi>=1.0->soundfile) (2.21)\n",
|
| 77 |
+
"Requirement already satisfied: aiohappyeyeballs>=2.3.0 in /opt/conda/lib/python3.10/site-packages (from aiohttp->datasets[audio]) (2.4.4)\n",
|
| 78 |
+
"Requirement already satisfied: aiosignal>=1.1.2 in /opt/conda/lib/python3.10/site-packages (from aiohttp->datasets[audio]) (1.3.1)\n",
|
| 79 |
+
"Requirement already satisfied: async-timeout<6.0,>=4.0 in /opt/conda/lib/python3.10/site-packages (from aiohttp->datasets[audio]) (5.0.1)\n",
|
| 80 |
+
"Requirement already satisfied: attrs>=17.3.0 in /opt/conda/lib/python3.10/site-packages (from aiohttp->datasets[audio]) (23.1.0)\n",
|
| 81 |
+
"Requirement already satisfied: frozenlist>=1.1.1 in /opt/conda/lib/python3.10/site-packages (from aiohttp->datasets[audio]) (1.5.0)\n",
|
| 82 |
+
"Requirement already satisfied: multidict<7.0,>=4.5 in /opt/conda/lib/python3.10/site-packages (from aiohttp->datasets[audio]) (6.1.0)\n",
|
| 83 |
+
"Requirement already satisfied: propcache>=0.2.0 in /opt/conda/lib/python3.10/site-packages (from aiohttp->datasets[audio]) (0.2.1)\n",
|
| 84 |
+
"Requirement already satisfied: yarl<2.0,>=1.17.0 in /opt/conda/lib/python3.10/site-packages (from aiohttp->datasets[audio]) (1.18.3)\n",
|
| 85 |
+
"Requirement already satisfied: jedi>=0.16 in /opt/conda/lib/python3.10/site-packages (from ipython>=6.1.0->ipywidgets) (0.18.1)\n",
|
| 86 |
+
"Requirement already satisfied: matplotlib-inline in /opt/conda/lib/python3.10/site-packages (from ipython>=6.1.0->ipywidgets) (0.1.6)\n",
|
| 87 |
+
"Requirement already satisfied: prompt-toolkit<3.1.0,>=3.0.41 in /opt/conda/lib/python3.10/site-packages (from ipython>=6.1.0->ipywidgets) (3.0.43)\n",
|
| 88 |
+
"Requirement already satisfied: pygments>=2.4.0 in /opt/conda/lib/python3.10/site-packages (from ipython>=6.1.0->ipywidgets) (2.15.1)\n",
|
| 89 |
+
"Requirement already satisfied: stack-data in /opt/conda/lib/python3.10/site-packages (from ipython>=6.1.0->ipywidgets) (0.2.0)\n",
|
| 90 |
+
"Requirement already satisfied: exceptiongroup in /opt/conda/lib/python3.10/site-packages (from ipython>=6.1.0->ipywidgets) (1.2.0)\n",
|
| 91 |
+
"Requirement already satisfied: pexpect>4.3 in /opt/conda/lib/python3.10/site-packages (from ipython>=6.1.0->ipywidgets) (4.8.0)\n",
|
| 92 |
+
"Requirement already satisfied: llvmlite<0.44,>=0.43.0dev0 in /opt/conda/lib/python3.10/site-packages (from numba>=0.51.0->librosa) (0.43.0)\n",
|
| 93 |
+
"Requirement already satisfied: platformdirs>=2.5.0 in /opt/conda/lib/python3.10/site-packages (from pooch>=1.1->librosa) (3.10.0)\n",
|
| 94 |
+
"Requirement already satisfied: charset-normalizer<4,>=2 in /opt/conda/lib/python3.10/site-packages (from requests>=2.32.2->datasets[audio]) (2.0.4)\n",
|
| 95 |
+
"Requirement already satisfied: idna<4,>=2.5 in /opt/conda/lib/python3.10/site-packages (from requests>=2.32.2->datasets[audio]) (3.4)\n",
|
| 96 |
+
"Requirement already satisfied: urllib3<3,>=1.21.1 in /opt/conda/lib/python3.10/site-packages (from requests>=2.32.2->datasets[audio]) (1.26.18)\n",
|
| 97 |
+
"Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.10/site-packages (from requests>=2.32.2->datasets[audio]) (2023.11.17)\n",
|
| 98 |
+
"Requirement already satisfied: threadpoolctl>=3.1.0 in /opt/conda/lib/python3.10/site-packages (from scikit-learn>=0.20.0->librosa) (3.5.0)\n",
|
| 99 |
+
"Requirement already satisfied: sympy in /opt/conda/lib/python3.10/site-packages (from torch->torchaudio) (1.12)\n",
|
| 100 |
+
"Requirement already satisfied: networkx in /opt/conda/lib/python3.10/site-packages (from torch->torchaudio) (3.1)\n",
|
| 101 |
+
"Requirement already satisfied: jinja2 in /opt/conda/lib/python3.10/site-packages (from torch->torchaudio) (3.1.2)\n",
|
| 102 |
+
"Requirement already satisfied: MarkupSafe>=2.1.1 in /opt/conda/lib/python3.10/site-packages (from werkzeug>=1.0.1->tensorboard) (2.1.3)\n",
|
| 103 |
+
"Requirement already satisfied: pytz>=2020.1 in /opt/conda/lib/python3.10/site-packages (from pandas->datasets[audio]) (2023.3.post1)\n",
|
| 104 |
+
"Requirement already satisfied: tzdata>=2022.7 in /opt/conda/lib/python3.10/site-packages (from pandas->datasets[audio]) (2024.2)\n",
|
| 105 |
+
"Requirement already satisfied: parso<0.9.0,>=0.8.0 in /opt/conda/lib/python3.10/site-packages (from jedi>=0.16->ipython>=6.1.0->ipywidgets) (0.8.3)\n",
|
| 106 |
+
"Requirement already satisfied: ptyprocess>=0.5 in /opt/conda/lib/python3.10/site-packages (from pexpect>4.3->ipython>=6.1.0->ipywidgets) (0.7.0)\n",
|
| 107 |
+
"Requirement already satisfied: wcwidth in /opt/conda/lib/python3.10/site-packages (from prompt-toolkit<3.1.0,>=3.0.41->ipython>=6.1.0->ipywidgets) (0.2.5)\n",
|
| 108 |
+
"Requirement already satisfied: executing in /opt/conda/lib/python3.10/site-packages (from stack-data->ipython>=6.1.0->ipywidgets) (0.8.3)\n",
|
| 109 |
+
"Requirement already satisfied: asttokens in /opt/conda/lib/python3.10/site-packages (from stack-data->ipython>=6.1.0->ipywidgets) (2.0.5)\n",
|
| 110 |
+
"Requirement already satisfied: pure-eval in /opt/conda/lib/python3.10/site-packages (from stack-data->ipython>=6.1.0->ipywidgets) (0.2.2)\n",
|
| 111 |
+
"Requirement already satisfied: mpmath>=0.19 in /opt/conda/lib/python3.10/site-packages (from sympy->torch->torchaudio) (1.3.0)\n",
|
| 112 |
+
"\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n",
|
| 113 |
+
"\u001b[0mNote: you may need to restart the kernel to use updated packages.\n"
|
| 114 |
+
]
|
| 115 |
+
}
|
| 116 |
+
],
|
| 117 |
+
"source": [
|
| 118 |
+
"%pip install librosa soundfile torchaudio datasets[audio] transformers[torch] audiomentations evaluate ipywidgets matplotlib tensorboard"
|
| 119 |
+
]
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"cell_type": "code",
|
| 123 |
+
"execution_count": 2,
|
| 124 |
+
"metadata": {},
|
| 125 |
+
"outputs": [],
|
| 126 |
+
"source": [
|
| 127 |
+
"import torch\n",
|
| 128 |
+
"import torchaudio\n",
|
| 129 |
+
"import librosa\n",
|
| 130 |
+
"import soundfile as sf\n",
|
| 131 |
+
"import numpy as np\n",
|
| 132 |
+
"import os\n",
|
| 133 |
+
"import matplotlib.pyplot as plt\n",
|
| 134 |
+
"import IPython.display as ipd\n",
|
| 135 |
+
"import datasets\n",
|
| 136 |
+
"import evaluate\n",
|
| 137 |
+
"from concurrent.futures import ProcessPoolExecutor\n",
|
| 138 |
+
"from transformers import ASTForAudioClassification, ASTFeatureExtractor, ASTConfig, TrainingArguments, Trainer\n",
|
| 139 |
+
"from audiomentations import Compose, AddGaussianSNR, GainTransition, Gain, ClippingDistortion, TimeStretch, PitchShift\n",
|
| 140 |
+
"from tqdm import tqdm"
|
| 141 |
+
]
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"cell_type": "code",
|
| 145 |
+
"execution_count": 3,
|
| 146 |
+
"metadata": {},
|
| 147 |
+
"outputs": [],
|
| 148 |
+
"source": [
|
| 149 |
+
"MODEL_DIR = \"/workspace\""
|
| 150 |
+
]
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"cell_type": "code",
|
| 154 |
+
"execution_count": 4,
|
| 155 |
+
"metadata": {},
|
| 156 |
+
"outputs": [
|
| 157 |
+
{
|
| 158 |
+
"data": {
|
| 159 |
+
"application/vnd.jupyter.widget-view+json": {
|
| 160 |
+
"model_id": "03839539e36849f181d0d21bdbf63073",
|
| 161 |
+
"version_major": 2,
|
| 162 |
+
"version_minor": 0
|
| 163 |
+
},
|
| 164 |
+
"text/plain": [
|
| 165 |
+
"Resolving data files: 0%| | 0/147 [00:00<?, ?it/s]"
|
| 166 |
+
]
|
| 167 |
+
},
|
| 168 |
+
"metadata": {},
|
| 169 |
+
"output_type": "display_data"
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"data": {
|
| 173 |
+
"application/vnd.jupyter.widget-view+json": {
|
| 174 |
+
"model_id": "cf03c7eeca6647fc8853e507ca878b03",
|
| 175 |
+
"version_major": 2,
|
| 176 |
+
"version_minor": 0
|
| 177 |
+
},
|
| 178 |
+
"text/plain": [
|
| 179 |
+
"Resolving data files: 0%| | 0/147 [00:00<?, ?it/s]"
|
| 180 |
+
]
|
| 181 |
+
},
|
| 182 |
+
"metadata": {},
|
| 183 |
+
"output_type": "display_data"
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"data": {
|
| 187 |
+
"application/vnd.jupyter.widget-view+json": {
|
| 188 |
+
"model_id": "147d4213469f4fd294bdd89a4f633f1b",
|
| 189 |
+
"version_major": 2,
|
| 190 |
+
"version_minor": 0
|
| 191 |
+
},
|
| 192 |
+
"text/plain": [
|
| 193 |
+
"Loading dataset shards: 0%| | 0/115 [00:00<?, ?it/s]"
|
| 194 |
+
]
|
| 195 |
+
},
|
| 196 |
+
"metadata": {},
|
| 197 |
+
"output_type": "display_data"
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"name": "stdout",
|
| 201 |
+
"output_type": "stream",
|
| 202 |
+
"text": [
|
| 203 |
+
"DatasetDict({\n",
|
| 204 |
+
" train: Dataset({\n",
|
| 205 |
+
" features: ['audio', 'source', 'ai_generated'],\n",
|
| 206 |
+
" num_rows: 20000\n",
|
| 207 |
+
" })\n",
|
| 208 |
+
"})\n"
|
| 209 |
+
]
|
| 210 |
+
}
|
| 211 |
+
],
|
| 212 |
+
"source": [
|
| 213 |
+
"# Load the dataset\n",
|
| 214 |
+
"ds = datasets.load_dataset(\"SleepyJesse/ai_music_large\")\n",
|
| 215 |
+
"# Resample the audio files to 16kHz\n",
|
| 216 |
+
"ds = ds.cast_column(\"audio\", datasets.Audio(sampling_rate=16000, mono=True))\n",
|
| 217 |
+
"print(ds)"
|
| 218 |
+
]
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"cell_type": "code",
|
| 222 |
+
"execution_count": 5,
|
| 223 |
+
"metadata": {
|
| 224 |
+
"editable": true,
|
| 225 |
+
"slideshow": {
|
| 226 |
+
"slide_type": ""
|
| 227 |
+
},
|
| 228 |
+
"tags": []
|
| 229 |
+
},
|
| 230 |
+
"outputs": [],
|
| 231 |
+
"source": [
|
| 232 |
+
"# Cast the \"ai_generated\" column (boolean) to class labels (\"ai_generated\" or \"human\")\n",
|
| 233 |
+
"class_labels = datasets.ClassLabel(names=[\"human\", \"ai_generated\"])\n",
|
| 234 |
+
"labels = [1 if x else 0 for x in ds['train']['ai_generated']]\n",
|
| 235 |
+
"ds['train'] = ds['train'].add_column(\"labels\", labels, feature=class_labels)"
|
| 236 |
+
]
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"cell_type": "code",
|
| 240 |
+
"execution_count": 6,
|
| 241 |
+
"metadata": {},
|
| 242 |
+
"outputs": [],
|
| 243 |
+
"source": [
|
| 244 |
+
"# Remove the \"ai_generated\" and \"source\" columns\n",
|
| 245 |
+
"ds[\"train\"] = ds[\"train\"].remove_columns(\"ai_generated\")\n",
|
| 246 |
+
"ds[\"train\"] = ds[\"train\"].remove_columns(\"source\")\n",
|
| 247 |
+
"\n",
|
| 248 |
+
"# Rename the \"audio\" column to \"input_values\" to match the expected input key for the processor\n",
|
| 249 |
+
"ds = ds.rename_column(\"audio\", \"input_values\")"
|
| 250 |
+
]
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"cell_type": "code",
|
| 254 |
+
"execution_count": 7,
|
| 255 |
+
"metadata": {},
|
| 256 |
+
"outputs": [
|
| 257 |
+
{
|
| 258 |
+
"name": "stdout",
|
| 259 |
+
"output_type": "stream",
|
| 260 |
+
"text": [
|
| 261 |
+
"DatasetDict({\n",
|
| 262 |
+
" train: Dataset({\n",
|
| 263 |
+
" features: ['input_values', 'labels'],\n",
|
| 264 |
+
" num_rows: 20000\n",
|
| 265 |
+
" })\n",
|
| 266 |
+
"})\n",
|
| 267 |
+
"{'input_values': {'path': '030312.mp3', 'array': array([ 0. , 0. , 0. , ..., -0.00048378,\n",
|
| 268 |
+
" -0.00049008, 0. ]), 'sampling_rate': 16000}, 'labels': 0}\n"
|
| 269 |
+
]
|
| 270 |
+
}
|
| 271 |
+
],
|
| 272 |
+
"source": [
|
| 273 |
+
"print(ds)\n",
|
| 274 |
+
"print(ds[\"train\"][0])"
|
| 275 |
+
]
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"cell_type": "code",
|
| 279 |
+
"execution_count": 8,
|
| 280 |
+
"metadata": {},
|
| 281 |
+
"outputs": [],
|
| 282 |
+
"source": [
|
| 283 |
+
"model_name = \"MIT/ast-finetuned-audioset-10-10-0.4593\" # Pre-trained AST model\n",
|
| 284 |
+
"feature_extractor = ASTFeatureExtractor.from_pretrained(model_name)\n",
|
| 285 |
+
"model_input_name = feature_extractor.model_input_names[0]\n",
|
| 286 |
+
"sampling_rate = feature_extractor.sampling_rate"
|
| 287 |
+
]
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"cell_type": "code",
|
| 291 |
+
"execution_count": 9,
|
| 292 |
+
"metadata": {},
|
| 293 |
+
"outputs": [],
|
| 294 |
+
"source": [
|
| 295 |
+
"# Define a function to preprocess the audio data\n",
|
| 296 |
+
"def preprocess_audio(batch):\n",
|
| 297 |
+
" wavs = [audio[\"array\"] for audio in batch[\"input_values\"]]\n",
|
| 298 |
+
" # inputs are spectrograms as torch.tensors now\n",
|
| 299 |
+
" inputs = feature_extractor(wavs, sampling_rate=sampling_rate, return_tensors=\"pt\")\n",
|
| 300 |
+
"\n",
|
| 301 |
+
" output_batch = {model_input_name: inputs.get(model_input_name), \"labels\": list(batch[\"labels\"])}\n",
|
| 302 |
+
" return output_batch\n",
|
| 303 |
+
"\n",
|
| 304 |
+
"# Apply the preprocessing function to the dataset\n",
|
| 305 |
+
"ds[\"train\"].set_transform(preprocess_audio, output_all_columns=False)"
|
| 306 |
+
]
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"cell_type": "code",
|
| 310 |
+
"execution_count": 10,
|
| 311 |
+
"metadata": {},
|
| 312 |
+
"outputs": [],
|
| 313 |
+
"source": [
|
| 314 |
+
"# Create audio augmentations\n",
|
| 315 |
+
"audio_augmentations = Compose([\n",
|
| 316 |
+
" AddGaussianSNR(min_snr_db=10, max_snr_db=20),\n",
|
| 317 |
+
" Gain(min_gain_db=-6, max_gain_db=6),\n",
|
| 318 |
+
" GainTransition(min_gain_db=-6, max_gain_db=6, min_duration=0.01, max_duration=0.3, duration_unit=\"fraction\"),\n",
|
| 319 |
+
" ClippingDistortion(min_percentile_threshold=0, max_percentile_threshold=30, p=0.5),\n",
|
| 320 |
+
" TimeStretch(min_rate=0.8, max_rate=1.2),\n",
|
| 321 |
+
" PitchShift(min_semitones=-4, max_semitones=4),\n",
|
| 322 |
+
"], p=0.5, shuffle=True)"
|
| 323 |
+
]
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"cell_type": "code",
|
| 327 |
+
"execution_count": 11,
|
| 328 |
+
"metadata": {},
|
| 329 |
+
"outputs": [],
|
| 330 |
+
"source": [
|
| 331 |
+
"# Define the preprocessing function for the audio augmentations\n",
|
| 332 |
+
"def preprocess_audio_with_transforms(batch):\n",
|
| 333 |
+
" # we apply augmentations on each waveform\n",
|
| 334 |
+
" wavs = [audio_augmentations(audio[\"array\"], sample_rate=sampling_rate) for audio in batch[\"input_values\"]]\n",
|
| 335 |
+
" inputs = feature_extractor(wavs, sampling_rate=sampling_rate, return_tensors=\"pt\")\n",
|
| 336 |
+
"\n",
|
| 337 |
+
" output_batch = {model_input_name: inputs.get(model_input_name), \"labels\": list(batch[\"labels\"])}\n",
|
| 338 |
+
" return output_batch"
|
| 339 |
+
]
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"cell_type": "code",
|
| 343 |
+
"execution_count": 12,
|
| 344 |
+
"metadata": {},
|
| 345 |
+
"outputs": [],
|
| 346 |
+
"source": [
|
| 347 |
+
"# Calculate values for normalization (mean and std) for the dataset (Only need to run this once per dataset)\n",
|
| 348 |
+
"# feature_extractor.do_normalize = False # Disable normalization\n",
|
| 349 |
+
"\n",
|
| 350 |
+
"# means = []\n",
|
| 351 |
+
"# stds = []\n",
|
| 352 |
+
"\n",
|
| 353 |
+
"# def calculate_mean_std(index):\n",
|
| 354 |
+
"# try:\n",
|
| 355 |
+
"# audio_input = ds[\"train\"][index][\"input_values\"]\n",
|
| 356 |
+
"# except Exception as e:\n",
|
| 357 |
+
"# print(f\"Error processing index {index}: {e}\")\n",
|
| 358 |
+
"# return None, None\n",
|
| 359 |
+
"# cur_mean = torch.mean(audio_input)\n",
|
| 360 |
+
"# cur_std = torch.std(audio_input)\n",
|
| 361 |
+
"# return cur_mean, cur_std\n",
|
| 362 |
+
"\n",
|
| 363 |
+
"# with ProcessPoolExecutor() as executor:\n",
|
| 364 |
+
"# results = list(tqdm(executor.map(calculate_mean_std, range(len(ds[\"train\"]))), total=len(ds[\"train\"])))\n",
|
| 365 |
+
"\n",
|
| 366 |
+
"# means, stds = zip(*results)\n",
|
| 367 |
+
"# means = [x.item() for x in means if x is not None]\n",
|
| 368 |
+
"# stds = [x.item() for x in stds if x is not None]\n",
|
| 369 |
+
"# feature_extractor.mean = torch.tensor(means).mean().item()\n",
|
| 370 |
+
"# feature_extractor.std = torch.tensor(stds).mean().item()\n",
|
| 371 |
+
"# feature_extractor.do_normalize = True # Enable normalization\n",
|
| 372 |
+
"\n",
|
| 373 |
+
"# print(f\"Mean: {feature_extractor.mean}\")\n",
|
| 374 |
+
"# print(f\"Std: {feature_extractor.std}\")\n",
|
| 375 |
+
"# print(\"Save these values for normalization if you're using the same dataset in the future.\")"
|
| 376 |
+
]
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"cell_type": "code",
|
| 380 |
+
"execution_count": 13,
|
| 381 |
+
"metadata": {},
|
| 382 |
+
"outputs": [],
|
| 383 |
+
"source": [
|
| 384 |
+
"# Remove corrupted audio files (4481, 8603 in ai_music_large)\n",
|
| 385 |
+
"corrupted_audio_indices = [4481, 8603]\n",
|
| 386 |
+
"keep_indices = [i for i in range(len(ds[\"train\"])) if i not in corrupted_audio_indices]\n",
|
| 387 |
+
"ds[\"train\"] = ds[\"train\"].select(keep_indices, writer_batch_size=50)"
|
| 388 |
+
]
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"cell_type": "code",
|
| 392 |
+
"execution_count": 14,
|
| 393 |
+
"metadata": {},
|
| 394 |
+
"outputs": [],
|
| 395 |
+
"source": [
|
| 396 |
+
"# Set the normalization values in the feature extractor (the following values are for the ai_music_large dataset)\n",
|
| 397 |
+
"feature_extractor.mean = -4.855465888977051\n",
|
| 398 |
+
"feature_extractor.std = 3.2848217487335205"
|
| 399 |
+
]
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"cell_type": "code",
|
| 403 |
+
"execution_count": 15,
|
| 404 |
+
"metadata": {},
|
| 405 |
+
"outputs": [
|
| 406 |
+
{
|
| 407 |
+
"name": "stdout",
|
| 408 |
+
"output_type": "stream",
|
| 409 |
+
"text": [
|
| 410 |
+
"Mean: -4.855465888977051\n",
|
| 411 |
+
"Std: 3.2848217487335205\n"
|
| 412 |
+
]
|
| 413 |
+
}
|
| 414 |
+
],
|
| 415 |
+
"source": [
|
| 416 |
+
"print(f\"Mean: {feature_extractor.mean}\")\n",
|
| 417 |
+
"print(f\"Std: {feature_extractor.std}\")"
|
| 418 |
+
]
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"cell_type": "code",
|
| 422 |
+
"execution_count": 16,
|
| 423 |
+
"metadata": {},
|
| 424 |
+
"outputs": [
|
| 425 |
+
{
|
| 426 |
+
"name": "stdout",
|
| 427 |
+
"output_type": "stream",
|
| 428 |
+
"text": [
|
| 429 |
+
"DatasetDict({\n",
|
| 430 |
+
" train: Dataset({\n",
|
| 431 |
+
" features: ['input_values', 'labels'],\n",
|
| 432 |
+
" num_rows: 15998\n",
|
| 433 |
+
" })\n",
|
| 434 |
+
" test: Dataset({\n",
|
| 435 |
+
" features: ['input_values', 'labels'],\n",
|
| 436 |
+
" num_rows: 4000\n",
|
| 437 |
+
" })\n",
|
| 438 |
+
"})\n"
|
| 439 |
+
]
|
| 440 |
+
}
|
| 441 |
+
],
|
| 442 |
+
"source": [
|
| 443 |
+
"# Split the dataset\n",
|
| 444 |
+
"if \"test\" not in ds:\n",
|
| 445 |
+
" ds = ds[\"train\"].train_test_split(test_size=0.2, shuffle=True, seed=42, stratify_by_column=\"labels\")\n",
|
| 446 |
+
"\n",
|
| 447 |
+
"print(ds)"
|
| 448 |
+
]
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"cell_type": "code",
|
| 452 |
+
"execution_count": 17,
|
| 453 |
+
"metadata": {},
|
| 454 |
+
"outputs": [],
|
| 455 |
+
"source": [
|
| 456 |
+
"# Set transforms for the train and test sets\n",
|
| 457 |
+
"ds[\"train\"].set_transform(preprocess_audio_with_transforms, output_all_columns=False)\n",
|
| 458 |
+
"ds[\"test\"].set_transform(preprocess_audio, output_all_columns=False)"
|
| 459 |
+
]
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"cell_type": "code",
|
| 463 |
+
"execution_count": 18,
|
| 464 |
+
"metadata": {},
|
| 465 |
+
"outputs": [
|
| 466 |
+
{
|
| 467 |
+
"name": "stderr",
|
| 468 |
+
"output_type": "stream",
|
| 469 |
+
"text": [
|
| 470 |
+
"Some weights of ASTForAudioClassification were not initialized from the model checkpoint at MIT/ast-finetuned-audioset-10-10-0.4593 and are newly initialized because the shapes did not match:\n",
|
| 471 |
+
"- classifier.dense.bias: found shape torch.Size([527]) in the checkpoint and torch.Size([2]) in the model instantiated\n",
|
| 472 |
+
"- classifier.dense.weight: found shape torch.Size([527, 768]) in the checkpoint and torch.Size([2, 768]) in the model instantiated\n",
|
| 473 |
+
"You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
|
| 474 |
+
]
|
| 475 |
+
}
|
| 476 |
+
],
|
| 477 |
+
"source": [
|
| 478 |
+
"# Load config from the pre-trained model\n",
|
| 479 |
+
"config = ASTConfig.from_pretrained(model_name)\n",
|
| 480 |
+
"\n",
|
| 481 |
+
"# Update the config with the labels we have in the dataset\n",
|
| 482 |
+
"config.num_labels = len(ds[\"train\"].features[\"labels\"].names)\n",
|
| 483 |
+
"config.label2id = {name: id for id, name in enumerate(ds[\"train\"].features[\"labels\"].names)}\n",
|
| 484 |
+
"config.id2label = {id: name for name, id in config.label2id.items()}\n",
|
| 485 |
+
"\n",
|
| 486 |
+
"# Initialize the model\n",
|
| 487 |
+
"model = ASTForAudioClassification.from_pretrained(model_name, config=config, ignore_mismatched_sizes=True)\n",
|
| 488 |
+
"model.init_weights()"
|
| 489 |
+
]
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"cell_type": "code",
|
| 493 |
+
"execution_count": 19,
|
| 494 |
+
"metadata": {},
|
| 495 |
+
"outputs": [],
|
| 496 |
+
"source": [
|
| 497 |
+
"# Configure the training arguments\n",
|
| 498 |
+
"training_args = TrainingArguments(\n",
|
| 499 |
+
" output_dir=MODEL_DIR + \"/out/ast_classifier_small\",\n",
|
| 500 |
+
" logging_dir=MODEL_DIR + \"/logs/ast_classifier_small\",\n",
|
| 501 |
+
" report_to=\"tensorboard\",\n",
|
| 502 |
+
" learning_rate=5e-5,\n",
|
| 503 |
+
" push_to_hub=False,\n",
|
| 504 |
+
" num_train_epochs=10,\n",
|
| 505 |
+
" per_device_train_batch_size=8,\n",
|
| 506 |
+
" eval_strategy=\"epoch\",\n",
|
| 507 |
+
" save_strategy=\"epoch\",\n",
|
| 508 |
+
" eval_steps=1,\n",
|
| 509 |
+
" save_steps=1,\n",
|
| 510 |
+
" logging_steps=10,\n",
|
| 511 |
+
" metric_for_best_model=\"accuracy\",\n",
|
| 512 |
+
" dataloader_num_workers=24,\n",
|
| 513 |
+
" dataloader_prefetch_factor=4,\n",
|
| 514 |
+
" dataloader_persistent_workers=True,\n",
|
| 515 |
+
")"
|
| 516 |
+
]
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"cell_type": "code",
|
| 520 |
+
"execution_count": 20,
|
| 521 |
+
"metadata": {},
|
| 522 |
+
"outputs": [],
|
| 523 |
+
"source": [
|
| 524 |
+
"# Define evaluation metrics\n",
|
| 525 |
+
"accuracy = evaluate.load(\"accuracy\")\n",
|
| 526 |
+
"recall = evaluate.load(\"recall\")\n",
|
| 527 |
+
"precision = evaluate.load(\"precision\")\n",
|
| 528 |
+
"f1 = evaluate.load(\"f1\")\n",
|
| 529 |
+
"\n",
|
| 530 |
+
"average = \"macro\" if config.num_labels > 2 else \"binary\"\n",
|
| 531 |
+
"\n",
|
| 532 |
+
"def compute_metrics(eval_pred):\n",
|
| 533 |
+
" logits = eval_pred.predictions\n",
|
| 534 |
+
" predictions = np.argmax(logits, axis=-1)\n",
|
| 535 |
+
" metrics = accuracy.compute(predictions=predictions, references=eval_pred.label_ids)\n",
|
| 536 |
+
" metrics.update(precision.compute(predictions=predictions, references=eval_pred.label_ids, average=average))\n",
|
| 537 |
+
" metrics.update(recall.compute(predictions=predictions, references=eval_pred.label_ids, average=average))\n",
|
| 538 |
+
" metrics.update(f1.compute(predictions=predictions, references=eval_pred.label_ids, average=average))\n",
|
| 539 |
+
" return metrics"
|
| 540 |
+
]
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"cell_type": "code",
|
| 544 |
+
"execution_count": 21,
|
| 545 |
+
"metadata": {},
|
| 546 |
+
"outputs": [
|
| 547 |
+
{
|
| 548 |
+
"name": "stderr",
|
| 549 |
+
"output_type": "stream",
|
| 550 |
+
"text": [
|
| 551 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 552 |
+
" warnings.warn(\n",
|
| 553 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 554 |
+
" warnings.warn(\n",
|
| 555 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 556 |
+
" warnings.warn(\n",
|
| 557 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 558 |
+
" warnings.warn(\n",
|
| 559 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 560 |
+
" warnings.warn(\n",
|
| 561 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 562 |
+
" warnings.warn(\n",
|
| 563 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 564 |
+
" warnings.warn(\n",
|
| 565 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 566 |
+
" warnings.warn(\n",
|
| 567 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 568 |
+
" warnings.warn(\n",
|
| 569 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 570 |
+
" warnings.warn(\n",
|
| 571 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 572 |
+
" warnings.warn(\n",
|
| 573 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 574 |
+
" warnings.warn(\n",
|
| 575 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 576 |
+
" warnings.warn(\n",
|
| 577 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 578 |
+
" warnings.warn(\n",
|
| 579 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 580 |
+
" warnings.warn(\n",
|
| 581 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 582 |
+
" warnings.warn(\n",
|
| 583 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 584 |
+
" warnings.warn(\n",
|
| 585 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 586 |
+
" warnings.warn(\n",
|
| 587 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 588 |
+
" warnings.warn(\n",
|
| 589 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 590 |
+
" warnings.warn(\n",
|
| 591 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 592 |
+
" warnings.warn(\n",
|
| 593 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 594 |
+
" warnings.warn(\n",
|
| 595 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 596 |
+
" warnings.warn(\n",
|
| 597 |
+
"/opt/conda/lib/python3.10/site-packages/audiomentations/core/transforms_interface.py:62: UserWarning: Warning: input samples dtype is np.float64. Converting to np.float32\n",
|
| 598 |
+
" warnings.warn(\n"
|
| 599 |
+
]
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"data": {
|
| 603 |
+
"text/html": [
|
| 604 |
+
"\n",
|
| 605 |
+
" <div>\n",
|
| 606 |
+
" \n",
|
| 607 |
+
" <progress value='20000' max='20000' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
|
| 608 |
+
" [20000/20000 2:46:13, Epoch 10/10]\n",
|
| 609 |
+
" </div>\n",
|
| 610 |
+
" <table border=\"1\" class=\"dataframe\">\n",
|
| 611 |
+
" <thead>\n",
|
| 612 |
+
" <tr style=\"text-align: left;\">\n",
|
| 613 |
+
" <th>Epoch</th>\n",
|
| 614 |
+
" <th>Training Loss</th>\n",
|
| 615 |
+
" <th>Validation Loss</th>\n",
|
| 616 |
+
" <th>Accuracy</th>\n",
|
| 617 |
+
" <th>Precision</th>\n",
|
| 618 |
+
" <th>Recall</th>\n",
|
| 619 |
+
" <th>F1</th>\n",
|
| 620 |
+
" </tr>\n",
|
| 621 |
+
" </thead>\n",
|
| 622 |
+
" <tbody>\n",
|
| 623 |
+
" <tr>\n",
|
| 624 |
+
" <td>5</td>\n",
|
| 625 |
+
" <td>0.031900</td>\n",
|
| 626 |
+
" <td>0.066149</td>\n",
|
| 627 |
+
" <td>0.982250</td>\n",
|
| 628 |
+
" <td>0.999482</td>\n",
|
| 629 |
+
" <td>0.965000</td>\n",
|
| 630 |
+
" <td>0.981938</td>\n",
|
| 631 |
+
" </tr>\n",
|
| 632 |
+
" <tr>\n",
|
| 633 |
+
" <td>6</td>\n",
|
| 634 |
+
" <td>0.234200</td>\n",
|
| 635 |
+
" <td>0.031733</td>\n",
|
| 636 |
+
" <td>0.992000</td>\n",
|
| 637 |
+
" <td>0.992000</td>\n",
|
| 638 |
+
" <td>0.992000</td>\n",
|
| 639 |
+
" <td>0.992000</td>\n",
|
| 640 |
+
" </tr>\n",
|
| 641 |
+
" <tr>\n",
|
| 642 |
+
" <td>7</td>\n",
|
| 643 |
+
" <td>0.063600</td>\n",
|
| 644 |
+
" <td>0.046821</td>\n",
|
| 645 |
+
" <td>0.992500</td>\n",
|
| 646 |
+
" <td>0.998987</td>\n",
|
| 647 |
+
" <td>0.986000</td>\n",
|
| 648 |
+
" <td>0.992451</td>\n",
|
| 649 |
+
" </tr>\n",
|
| 650 |
+
" <tr>\n",
|
| 651 |
+
" <td>8</td>\n",
|
| 652 |
+
" <td>0.210500</td>\n",
|
| 653 |
+
" <td>0.017158</td>\n",
|
| 654 |
+
" <td>0.995500</td>\n",
|
| 655 |
+
" <td>0.997990</td>\n",
|
| 656 |
+
" <td>0.993000</td>\n",
|
| 657 |
+
" <td>0.995489</td>\n",
|
| 658 |
+
" </tr>\n",
|
| 659 |
+
" <tr>\n",
|
| 660 |
+
" <td>9</td>\n",
|
| 661 |
+
" <td>0.001100</td>\n",
|
| 662 |
+
" <td>0.016046</td>\n",
|
| 663 |
+
" <td>0.996750</td>\n",
|
| 664 |
+
" <td>0.998995</td>\n",
|
| 665 |
+
" <td>0.994500</td>\n",
|
| 666 |
+
" <td>0.996743</td>\n",
|
| 667 |
+
" </tr>\n",
|
| 668 |
+
" <tr>\n",
|
| 669 |
+
" <td>10</td>\n",
|
| 670 |
+
" <td>0.001500</td>\n",
|
| 671 |
+
" <td>0.011154</td>\n",
|
| 672 |
+
" <td>0.997500</td>\n",
|
| 673 |
+
" <td>0.998497</td>\n",
|
| 674 |
+
" <td>0.996500</td>\n",
|
| 675 |
+
" <td>0.997497</td>\n",
|
| 676 |
+
" </tr>\n",
|
| 677 |
+
" </tbody>\n",
|
| 678 |
+
"</table><p>"
|
| 679 |
+
],
|
| 680 |
+
"text/plain": [
|
| 681 |
+
"<IPython.core.display.HTML object>"
|
| 682 |
+
]
|
| 683 |
+
},
|
| 684 |
+
"metadata": {},
|
| 685 |
+
"output_type": "display_data"
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"data": {
|
| 689 |
+
"text/plain": [
|
| 690 |
+
"TrainOutput(global_step=20000, training_loss=0.04969663131231209, metrics={'train_runtime': 9996.4827, 'train_samples_per_second': 16.004, 'train_steps_per_second': 2.001, 'total_flos': 1.084389872624468e+19, 'train_loss': 0.04969663131231209, 'epoch': 10.0})"
|
| 691 |
+
]
|
| 692 |
+
},
|
| 693 |
+
"execution_count": 21,
|
| 694 |
+
"metadata": {},
|
| 695 |
+
"output_type": "execute_result"
|
| 696 |
+
}
|
| 697 |
+
],
|
| 698 |
+
"source": [
|
| 699 |
+
"# Initialize the Trainer\n",
|
| 700 |
+
"trainer = Trainer(\n",
|
| 701 |
+
" model=model,\n",
|
| 702 |
+
" args=training_args,\n",
|
| 703 |
+
" train_dataset=ds[\"train\"],\n",
|
| 704 |
+
" eval_dataset=ds[\"test\"],\n",
|
| 705 |
+
" compute_metrics=compute_metrics,\n",
|
| 706 |
+
")\n",
|
| 707 |
+
"\n",
|
| 708 |
+
"# Train the model\n",
|
| 709 |
+
"trainer.train(resume_from_checkpoint=True)"
|
| 710 |
+
]
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"cell_type": "code",
|
| 714 |
+
"execution_count": 22,
|
| 715 |
+
"metadata": {},
|
| 716 |
+
"outputs": [],
|
| 717 |
+
"source": [
|
| 718 |
+
"trainer.save_model(output_dir=\"./model\")"
|
| 719 |
+
]
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"cell_type": "code",
|
| 723 |
+
"execution_count": 26,
|
| 724 |
+
"metadata": {},
|
| 725 |
+
"outputs": [
|
| 726 |
+
{
|
| 727 |
+
"name": "stdout",
|
| 728 |
+
"output_type": "stream",
|
| 729 |
+
"text": [
|
| 730 |
+
"Dataset({\n",
|
| 731 |
+
" features: ['input_values', 'labels'],\n",
|
| 732 |
+
" num_rows: 22\n",
|
| 733 |
+
"})\n"
|
| 734 |
+
]
|
| 735 |
+
}
|
| 736 |
+
],
|
| 737 |
+
"source": [
|
| 738 |
+
"import glob\n",
|
| 739 |
+
"unseen_files = glob.glob(\"/workspace/ai/*\")\n",
|
| 740 |
+
"unseen_set = datasets.Dataset.from_dict({\"input_values\": unseen_files}).cast_column(\"input_values\", datasets.Audio(sampling_rate=16000, mono=True))\n",
|
| 741 |
+
"unseen_set = unseen_set.add_column(name=\"labels\", column=[1 for _ in range(len(unseen_set))])\n",
|
| 742 |
+
"unseen_set.set_transform(preprocess_audio, output_all_columns=False)\n",
|
| 743 |
+
"print(unseen_set)"
|
| 744 |
+
]
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"cell_type": "code",
|
| 748 |
+
"execution_count": 27,
|
| 749 |
+
"metadata": {},
|
| 750 |
+
"outputs": [
|
| 751 |
+
{
|
| 752 |
+
"data": {
|
| 753 |
+
"text/html": [],
|
| 754 |
+
"text/plain": [
|
| 755 |
+
"<IPython.core.display.HTML object>"
|
| 756 |
+
]
|
| 757 |
+
},
|
| 758 |
+
"metadata": {},
|
| 759 |
+
"output_type": "display_data"
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"data": {
|
| 763 |
+
"text/plain": [
|
| 764 |
+
"PredictionOutput(predictions=array([[-6.1512766, 6.060689 ],\n",
|
| 765 |
+
" [-5.978138 , 5.743587 ],\n",
|
| 766 |
+
" [-4.0873733, 4.6713266],\n",
|
| 767 |
+
" [-4.008548 , 4.2211466],\n",
|
| 768 |
+
" [-5.873764 , 5.7459254],\n",
|
| 769 |
+
" [-6.206414 , 6.235821 ],\n",
|
| 770 |
+
" [-4.825156 , 4.8879967],\n",
|
| 771 |
+
" [ 2.4498227, -1.9184169],\n",
|
| 772 |
+
" [-5.554337 , 5.638381 ],\n",
|
| 773 |
+
" [-6.2935424, 6.2818317],\n",
|
| 774 |
+
" [-5.4350233, 5.3958435],\n",
|
| 775 |
+
" [-5.253522 , 5.241722 ],\n",
|
| 776 |
+
" [-3.9684274, 3.9555552],\n",
|
| 777 |
+
" [-6.3393865, 6.066998 ],\n",
|
| 778 |
+
" [-6.2268295, 5.997632 ],\n",
|
| 779 |
+
" [-6.1494975, 6.1331954],\n",
|
| 780 |
+
" [-5.7538185, 5.824904 ],\n",
|
| 781 |
+
" [ 3.1460629, -2.850086 ],\n",
|
| 782 |
+
" [-1.4815819, 2.1283977],\n",
|
| 783 |
+
" [-5.2852707, 5.146372 ],\n",
|
| 784 |
+
" [-6.6310973, 6.3678217],\n",
|
| 785 |
+
" [ 4.265127 , -3.486055 ]], dtype=float32), label_ids=array([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]), metrics={'test_loss': 0.8253983855247498, 'test_accuracy': 0.8636363636363636, 'test_precision': 1.0, 'test_recall': 0.8636363636363636, 'test_f1': 0.926829268292683, 'test_runtime': 6.5504, 'test_samples_per_second': 3.359, 'test_steps_per_second': 0.458})"
|
| 786 |
+
]
|
| 787 |
+
},
|
| 788 |
+
"execution_count": 27,
|
| 789 |
+
"metadata": {},
|
| 790 |
+
"output_type": "execute_result"
|
| 791 |
+
}
|
| 792 |
+
],
|
| 793 |
+
"source": [
|
| 794 |
+
"trainer.predict(unseen_set)"
|
| 795 |
+
]
|
| 796 |
+
}
|
| 797 |
+
],
|
| 798 |
+
"metadata": {
|
| 799 |
+
"kernelspec": {
|
| 800 |
+
"display_name": "base",
|
| 801 |
+
"language": "python",
|
| 802 |
+
"name": "python3"
|
| 803 |
+
},
|
| 804 |
+
"language_info": {
|
| 805 |
+
"codemirror_mode": {
|
| 806 |
+
"name": "ipython",
|
| 807 |
+
"version": 3
|
| 808 |
+
},
|
| 809 |
+
"file_extension": ".py",
|
| 810 |
+
"mimetype": "text/x-python",
|
| 811 |
+
"name": "python",
|
| 812 |
+
"nbconvert_exporter": "python",
|
| 813 |
+
"pygments_lexer": "ipython3",
|
| 814 |
+
"version": "3.10.13"
|
| 815 |
+
}
|
| 816 |
+
},
|
| 817 |
+
"nbformat": 4,
|
| 818 |
+
"nbformat_minor": 4
|
| 819 |
+
}
|