generate_BirdVox-full-night.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "import pandas as pd\n",
    "\n",
    "\n",
    "data_dir = '/beegfs/vl1019/BirdVox_datasets/'\n",
    "icassp_dir = os.path.join(data_dir, 'BirdVox-full-night_icassp2018')\n",
    "in_csv_dir = os.path.join(icassp_dir, 'BirdVox-70k_annotations')\n",
    "in_flac_dir = os.path.join(icassp_dir, 'BirdVox-70k_full-audio')\n",
    "\n",
    "full_night_dir = os.path.join(data_dir, \"BirdVox-full-night\")\n",
    "out_csv_dir = os.path.join(full_night_dir, \"BirdVox-full-night_csv-annotations\")\n",
    "os.makedirs(out_csv_dir, exist_ok=True)\n",
    "out_flac_dir = os.path.join(full_night_dir, \"BirdVox-full-night_flac-audio\")\n",
    "os.makedirs(out_flac_dir, exist_ok=True)\n",
    "\n",
    "units = [\"unit\" + str(n).zfill(2) for n in [1, 2, 3, 5, 7, 10]]\n",
    "\n",
    "# Loop over units, i.e. sensors.\n",
    "for unit_str in units:\n",
    "    # Copy FLAC files.\n",
    "    in_flac_path = os.path.join(in_flac_dir, unit_str + \".flac\")\n",
    "    out_flac_path = os.path.join(out_flac_dir, \n",
    "        \"BirdVox-full-night_flac-audio_\" + unit_str + \".flac\")\n",
    "    os.system(\" \".join([\"mv\", in_flac_path, out_flac_path]))\n",
    "    \n",
    "    # Create simplified CSV files.\n",
    "    unit_name = unit_str + \".txt\"\n",
    "    unit_path = os.path.join(in_csv_dir, unit_name)\n",
    "    df_in = pd.read_csv(unit_path, '\\t')\n",
    "    if \"Calls\" in df_in:\n",
    "        df_in = df_in[df_in[\"Calls\"] != \"alarm\"]\n",
    "    center_times = 0.5 * (df_in[\"Begin Time (s)\"] + df_in[\"End Time (s)\"])\n",
    "    center_time_strs = [\"{:9.3f}\".format(t) for t in center_times]\n",
    "\n",
    "    center_freqs = 0.5 * (df_in[\"Low Freq (Hz)\"] + df_in[\"High Freq (Hz)\"])\n",
    "    center_freq_strs = [\"{:5d}\".format(int(np.round(f))) for f in center_freqs]\n",
    "\n",
    "    df_out = pd.DataFrame({\"Time (s)\": center_time_strs, \"Freq (Hz)\": center_freq_strs},\n",
    "        columns=[\"Time (s)\", \"Freq (Hz)\"])\n",
    "    out_csv_path = os.path.join(out_csv_dir,\n",
    "        \"BirdVox-full-night_csv-annotations_\" + unit_str + \".csv\")\n",
    "    df_out.to_csv(out_csv_path, index=False)\n",
    "    \n",
    "    \n",
    "# I mixed latitude and longitude in the preliminary version of the dataset.\n",
    "# Here is a fix.\n",
    "in_gps_path = os.path.join(icassp_dir, \"BirdVox-70k_gps-coordinates.csv\")\n",
    "df_gps_in = pd.read_csv(in_gps_path)\n",
    "df_gps_out = pd.DataFrame({\n",
    "    \"Unit\": units,\n",
    "    \"Latitude\": df_gps_in[\"Longitude\"],\n",
    "    \"Longitude\": df_gps_in[\"Latitude\"]},\n",
    "    columns = [\"Unit\", \"Latitude\", \"Longitude\"])\n",
    "out_gps_path = os.path.join(full_night_dir, \"BirdVox-full-night_gps-coordinates.csv\")\n",
    "df_gps_out.to_csv(out_gps_path, index=False)\n",
    "\n",
    "\n",
    "# Copy UTC timestamps.\n",
    "in_utc_path = os.path.join(icassp_dir, \"BirdVox-70k_utc-start-times.csv\")\n",
    "out_utc_path = os.path.join(full_night_dir, \n",
    "    \"BirdVox-full-night_utc-start-times.csv\")\n",
    "os.system(\" \".join([\"mv\", in_utc_path, out_utc_path]));"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}