m@6: {
m@6:  "cells": [
m@6:   {
m@6:    "cell_type": "code",
m@6:    "execution_count": 1,
m@6:    "metadata": {
m@6:     "collapsed": true
m@6:    },
m@6:    "outputs": [],
m@6:    "source": [
m@6:     "import numpy as np\n",
m@6:     "import pickle"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "code",
m@6:    "execution_count": 2,
m@6:    "metadata": {
m@6:     "collapsed": true
m@6:    },
m@6:    "outputs": [],
m@6:    "source": [
m@7:     "filenames = ['/import/c4dm-04/mariap/train_data_melodia_8.pickle',\n",
m@6:     "             '/import/c4dm-04/mariap/val_data_melodia_8.pickle', \n",
m@6:     "             '/import/c4dm-04/mariap/test_data_melodia_8.pickle']"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "code",
m@7:    "execution_count": 13,
Maria@40:    "metadata": {
Maria@40:     "collapsed": true
Maria@40:    },
m@6:    "outputs": [],
m@6:    "source": [
m@6:     "all_Yaudio = []\n",
m@7:     "all_Y = []\n",
m@6:     "for filename in filenames:\n",
m@7:     "    _, Y, Yaudio = pickle.load(open(filename, 'rb'))\n",
m@6:     "    all_Yaudio.append(Yaudio)\n",
m@7:     "    all_Y.append(Y)\n",
m@7:     "all_Yaudio = np.concatenate(all_Yaudio)\n",
m@7:     "all_Y = np.concatenate(all_Y)"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "code",
m@7:    "execution_count": 5,
Maria@40:    "metadata": {
Maria@40:     "collapsed": true
Maria@40:    },
m@7:    "outputs": [],
m@6:    "source": [
m@6:     "uniq_audio, uniq_counts = np.unique(all_Yaudio, return_counts=True)"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "markdown",
m@6:    "metadata": {},
m@6:    "source": [
m@6:     "## Stats on audio files with very few music frames (after the speech/music discrimination)"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "code",
m@7:    "execution_count": 8,
Maria@40:    "metadata": {
Maria@40:     "collapsed": false
Maria@40:    },
m@6:    "outputs": [
m@6:     {
m@7:      "name": "stdout",
m@7:      "output_type": "stream",
m@7:      "text": [
m@7:       "63 files out of 8200 have less than 10 frames\n"
m@6:      ]
m@6:     }
m@6:    ],
m@6:    "source": [
m@6:     "min_n_frames = 10\n",
m@7:     "short_files_idx = np.where(uniq_counts<min_n_frames)[0]\n",
m@7:     "print '%d files out of %d have less than %d frames' % (len(short_files_idx), len(uniq_counts), min_n_frames)"
m@7:    ]
m@7:   },
m@7:   {
m@7:    "cell_type": "markdown",
m@7:    "metadata": {},
m@7:    "source": [
m@7:     "Countries for tracks with less than 10 frames"
m@7:    ]
m@7:   },
m@7:   {
m@7:    "cell_type": "code",
m@7:    "execution_count": 17,
Maria@40:    "metadata": {
Maria@40:     "collapsed": false
Maria@40:    },
m@7:    "outputs": [
m@7:     {
m@7:      "name": "stdout",
m@7:      "output_type": "stream",
m@7:      "text": [
m@7:       "{'Italy': 1, 'Peru': 1, 'Solomon Islands': 2, 'France': 2, 'Ethiopia': 1, 'Somalia': 1, 'Ireland': 1, 'Swaziland': 1, 'Argentina': 1, 'Norway': 1, 'Nigeria': 1, 'Algeria': 1, 'Germany': 1, 'Puerto Rico': 1, 'Dominican Republic': 1, 'Poland': 3, 'Spain': 2, 'Netherlands': 1, 'Uganda': 4, 'Western Sahara': 5, 'Gambia': 2, 'Philippines': 2, 'Trinidad and Tobago': 1, 'Latvia': 1, 'South Sudan': 3, 'Mali': 1, 'Russia': 1, 'Romania': 1, 'Portugal': 1, 'South Africa': 3, 'Egypt': 1, 'Sierra Leone': 1, 'United Kingdom': 4, 'Lesotho': 1, 'Senegal': 2, 'Colombia': 2, 'Japan': 2, 'Nicaragua': 1, 'Botswana': 1}\n"
m@7:      ]
m@7:     }
m@7:    ],
m@7:    "source": [
m@7:     "countries = np.array([all_Y[np.where(all_Yaudio==uniq_audio[audio_idx])[0][0]][0] for audio_idx in short_files_idx])\n",
m@7:     "unique, counts = np.unique(countries, return_counts=True)\n",
m@7:     "print dict(zip(unique, counts))"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "markdown",
m@6:    "metadata": {},
m@6:    "source": [
m@6:     "## Stats on average duration of the music segments for all tracks"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "code",
m@7:    "execution_count": 18,
Maria@40:    "metadata": {
Maria@40:     "collapsed": false
Maria@40:    },
m@6:    "outputs": [
m@6:     {
m@7:      "name": "stdout",
m@7:      "output_type": "stream",
m@7:      "text": [
m@7:       "mean 65.750000\n",
m@7:       "median 44.000000\n",
m@7:       "std 45.947865\n",
m@7:       "mean duration 32.875000\n"
m@6:      ]
m@6:     }
m@6:    ],
m@6:    "source": [
m@6:     "sr = 2.0  # with 8-second window and 0.5-second hop size the sampling rate is 2 about 2 samples per second\n",
m@6:     "print 'mean %f' % np.mean(uniq_counts)\n",
m@6:     "print 'median %f' % np.median(uniq_counts)\n",
m@6:     "print 'std %f' % np.std(uniq_counts)\n",
m@6:     "print 'mean duration %f' % (np.mean(uniq_counts) / sr)"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "markdown",
m@6:    "metadata": {},
m@6:    "source": [
m@7:     "## Stats on average duration of the music segments for the Smithsonian tracks"
m@7:    ]
m@7:   },
m@7:   {
m@7:    "cell_type": "code",
m@7:    "execution_count": 23,
Maria@40:    "metadata": {
Maria@40:     "collapsed": false
Maria@40:    },
m@7:    "outputs": [
m@7:     {
m@7:      "name": "stdout",
m@7:      "output_type": "stream",
m@7:      "text": [
m@7:       "n tracks: 6132\n",
m@7:       "mean 42.618885\n",
m@7:       "median 44.000000\n",
m@7:       "std 4.804534\n",
m@7:       "mean duration 21.309442\n"
m@7:      ]
m@7:     }
m@7:    ],
m@7:    "source": [
m@7:     "#British library tracks start with 'D:/Audio/...'\n",
m@7:     "idx_SM_tracks = np.array([i for i in range(len(uniq_audio)) if len(uniq_audio[i].split('D:/'))==1])\n",
m@7:     "sr = 2.0  # with 8-second window and 0.5-second hop size the sampling rate is 2 about 2 samples per second\n",
m@7:     "print 'n tracks: %d' % len(idx_SM_tracks)\n",
m@7:     "print 'mean %f' % np.mean(uniq_counts[idx_SM_tracks])\n",
m@7:     "print 'median %f' % np.median(uniq_counts[idx_SM_tracks])\n",
m@7:     "print 'std %f' % np.std(uniq_counts[idx_SM_tracks])\n",
m@7:     "print 'mean duration %f' % (np.mean(uniq_counts[idx_SM_tracks]) / sr)"
m@7:    ]
m@7:   },
m@7:   {
m@7:    "cell_type": "markdown",
m@7:    "metadata": {},
m@7:    "source": [
m@6:     "## Stats on average duration of the music segments for the British Library tracks"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "code",
m@7:    "execution_count": 22,
Maria@40:    "metadata": {
Maria@40:     "collapsed": false
Maria@40:    },
m@6:    "outputs": [
m@6:     {
m@7:      "name": "stdout",
m@7:      "output_type": "stream",
m@7:      "text": [
m@7:       "n tracks: 2068\n",
m@7:       "mean 134.338008\n",
m@7:       "median 163.000000\n",
m@7:       "std 44.855790\n",
m@7:       "mean duration 67.169004\n"
m@6:      ]
m@6:     }
m@6:    ],
m@6:    "source": [
m@6:     "#British library tracks start with 'D:/Audio/...'\n",
m@6:     "idx_BL_tracks = np.array([i for i in range(len(uniq_audio)) if len(uniq_audio[i].split('D:/'))>1])\n",
m@6:     "sr = 2.0  # with 8-second window and 0.5-second hop size the sampling rate is 2 about 2 samples per second\n",
m@7:     "print 'n tracks: %d' % len(idx_BL_tracks)\n",
m@6:     "print 'mean %f' % np.mean(uniq_counts[idx_BL_tracks])\n",
m@6:     "print 'median %f' % np.median(uniq_counts[idx_BL_tracks])\n",
m@6:     "print 'std %f' % np.std(uniq_counts[idx_BL_tracks])\n",
m@6:     "print 'mean duration %f' % (np.mean(uniq_counts[idx_BL_tracks]) / sr)"
m@6:    ]
m@6:   },
m@6:   {
m@6:    "cell_type": "code",
m@6:    "execution_count": null,
m@6:    "metadata": {
m@6:     "collapsed": true
m@6:    },
m@6:    "outputs": [],
m@6:    "source": []
m@6:   }
m@6:  ],
m@6:  "metadata": {
m@6:   "kernelspec": {
m@6:    "display_name": "Python 2",
m@6:    "language": "python",
m@6:    "name": "python2"
m@6:   },
m@6:   "language_info": {
m@6:    "codemirror_mode": {
m@6:     "name": "ipython",
m@6:     "version": 2
m@6:    },
m@6:    "file_extension": ".py",
m@6:    "mimetype": "text/x-python",
m@6:    "name": "python",
m@6:    "nbconvert_exporter": "python",
m@6:    "pygments_lexer": "ipython2",
Maria@40:    "version": "2.7.11"
m@6:   }
m@6:  },
m@6:  "nbformat": 4,
m@7:  "nbformat_minor": 1
m@6: }