create_audio_features.py.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import essentia\n",
    "import essentia.standard as es\n",
    "import glob\n",
    "import librosa\n",
    "import soundfile as sf\n",
    "from essentia.standard import *\n",
    "import numpy as np\n",
    "import ntpath\n",
    "import pandas as pd\n",
    "import logging\n",
    "from multiprocessing import Pool\n",
    "import traceback\n",
    "logging.basicConfig(filename='audio_feature_extraction.log',level=logging.DEBUG)\n",
    "\n",
    "import argparse\n",
    "\n",
    "parser = argparse.ArgumentParser(description='Script to create audio features')\n",
    "parser.add_argument('--in_wav_dirs', nargs='+',required=True, help='Video directories that contains wav files')\n",
    "parser.add_argument('--csv_path', type=str, default=\"csv/audio_features\", help='.csv file contain audio features')\n",
    "\n",
    "args = parser.parse_args()\n",
    "in_wav_dirs = args.in_wav_dirs\n",
    "csv_path = args.csv_path\n",
    "\n",
    "def get_audio_features(f):\n",
    "    features, features_frames = es.MusicExtractor(\n",
    "                                              lowlevelStats=['mean', 'stdev'],\n",
    "                                              rhythmStats=['mean', 'stdev'],\n",
    "                                              tonalStats=['mean', 'stdev'],\n",
    "                                             )(f)\n",
    "    features_name = features.descriptorNames()\n",
    "    features_name = list(filter(lambda name: \"metadata\" not in name,features_name))\n",
    "    info = {}\n",
    "    for f in features_name:\n",
    "        if not isinstance(features[f], np.ndarray):\n",
    "            info[f] = features[f]\n",
    "    \n",
    "    return info\n",
    "\n",
    "def task(file, ID):\n",
    "    try:\n",
    "        feature = get_audio_features(file)\n",
    "        logging.info(\"processed file {}\".format(file))\n",
    "        return (feature, ID)\n",
    "    except Exception as e:\n",
    "        logging.error(traceback.format_exc())\n",
    "        return (None, None)\n",
    "\n",
    "    \n",
    "def get_all_audio_features_parallel(list_dir, out_csv):\n",
    "    files = []\n",
    "    for directory in list_dir:\n",
    "        files += glob.glob(directory + \"/*.wav\")\n",
    "    \n",
    "    IDs = [ntpath.basename(f).replace(\".mp3\", \"\").replace(\".wav\", \"\") for f in files]\n",
    "    \n",
    "    with Pool(processes=7) as pool:\n",
    "        in_out = pool.starmap(task, list(zip(files, IDs)))\n",
    "\n",
    "        \n",
    "    in_out = [p  for p in in_out if p[0] is not None]\n",
    "    input, output = zip(*in_out)    \n",
    "    features = np.array(input)\n",
    "    IDs = np.array(output)\n",
    "    \n",
    "    keys = features[0].keys()\n",
    "    data = {}\n",
    "    data[\"ID\"] = IDs\n",
    "    for k in keys:\n",
    "        data[k] = [feature[k] for feature in features]\n",
    "        \n",
    "        \n",
    "    df = pd.DataFrame(data=data)\n",
    "    \n",
    "    df.to_csv(out_csv)\n",
    "    \n",
    "get_all_audio_features_parallel(list_dir=in_wav_dirs, out_csv=csv_path)\n",
    "# get_all_audio_features_parallel([\"/data/zalo/hit-song-prediction/train-wav-samples\", \"/data/zalo/hit-song-prediction/test-wav-samples\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (tf-gpu)",
   "language": "python",
   "name": "tf-gpu"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}