notes_recognition.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from librosa import load, logamplitude\n",
    "from librosa.feature import melspectrogram\n",
    "from tqdm import tqdm_notebook\n",
    "from multiprocessing import Pool\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import os\n",
    "filesNames = os.listdir(\"./wav\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def get_spectrogram(path):\n",
    "    \"\"\"Строим спектограмму из wav файла\"\"\"\n",
    "#   Кстати, поправочка, которая стоила мне нервов: librosa работает только с wav файлами с частотой дискретизации 44,1 кГц\n",
    "    y, sr = load(path)\n",
    "    S = melspectrogram(y, sr=sr, n_mels=100)\n",
    "    log_S = logamplitude(S, ref_power=np.max)\n",
    "    return log_S"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def get_spectrogram_1(x):\n",
    "#   Отправляем путь к файлу в get_spectrogram()\n",
    "    i, x = x\n",
    "    if i % 1000 == 0:\n",
    "        print(i)\n",
    "    return get_spectrogram('./wav/' + x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "spectrograms = []\n",
    "if os.stat(\"saved_data\").st_size != 0:\n",
    "    spectrograms = pickle.load(open(\"saved_data\", \"rb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if len(filesNames) != len(spectrograms):\n",
    "    p = Pool(4)\n",
    "    spectrograms = p.map(get_spectrogram_1, enumerate(a))\n",
    "    pickle.dump(spectrograms, open(\"saved_data\", \"wb\"))    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Создаем матрицу из нулей (кол-во_аудиофайлов, кол-во_нот_на_клавиатуре_ф-но_=_88)\n",
    "number_of_notes = 88\n",
    "data_answ = np.zeros((len(spectrograms), number_of_notes))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range (len(spectrograms)):\n",
    "#   Берем инфорацию из имени файла, записываем в матрицу единицы напротив звучащих в файле нот\n",
    "#   file name example 2,23,36,120.mid.wav\n",
    "    name = filesNames[i][:-8].split(\",\") \n",
    "    for j in range (int(name[0])):\n",
    "        data_answ[i][int(name[j+1]) - 21] = 1  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "data_answ = np.array(data_answ, dtype=int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(100, 90)\n",
      "(100, 95)\n",
      "(100, 145)\n"
     ]
    }
   ],
   "source": [
    "# У всех спектрограмм разное кол-во столбцов (и, возможно, строк)\n",
    "print (spectrograms[0].shape)\n",
    "print (spectrograms[1].shape)\n",
    "print (spectrograms[8000].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Приведем матрицы к наименьшему размеру, существующему в данной выборке\n",
    "arr = []\n",
    "min_num_of_rows = min(list(map(lambda x: np.shape(x)[0], spectrograms)))\n",
    "min_num_of_cols = min(list(map(lambda x: np.shape(x)[1], spectrograms)))\n",
    "for i in spectrograms:\n",
    "    arr.append(i[:min_num_of_rows, :min_num_of_cols])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Так как spectrograms - лист двумерных массивов, а модель принимает на вход только двумерные массивы, \n",
    "# преобразуем spectrograms, сделав из матрицы одномерный массив\n",
    "spectrograms = np.array([b.ravel() for b in arr])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(21181, 8900)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spectrograms.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# выберем достаточно большое рандомное число < 21181\n",
    "random_number = 15000\n",
    "X_train = spectrograms[:random_number]\n",
    "Y_train = data_answ[:random_number]\n",
    "X_test = spectrograms[random_number:]\n",
    "Y_test = data_answ[random_number:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from tqdm import tqdm_notebook"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "models = []\n",
    "if os.stat(\"Models_trained\").st_size != 0:\n",
    "    models = pickle.load(open(\"Models_trained\", \"rb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if (len(models) != number_of_notes):\n",
    "    for i in tqdm_notebook(range(number_of_notes)):\n",
    "        model = LogisticRegression()\n",
    "        model.fit(X_train, Y_train[:, i])\n",
    "        models.append(model)\n",
    "    pickle.dump(models, open(\"Models_trained\", \"wb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import roc_auc_score\n",
    "Y_pred_trains =[]\n",
    "Y_pred_tests = []\n",
    "for i in tqdm_notebook(range(88)):\n",
    "    Y_pred_trains.append(models[i].predict(X_train))\n",
    "    Y_pred_tests.append(models[i].predict(X_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f348abc86d8>]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhIAAAFkCAYAAAB1rtL+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f348abc8668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trains = []\n",
    "tests = []\n",
    "for i in tqdm_notebook(range(88)):\n",
    "    trains.append(roc_auc_score(Y_train[:, i],Y_pred_trains[i]))\n",
    "    tests.append(roc_auc_score(Y_test[:, i],Y_pred_tests[i]))\n",
    "plt.plot(np.arange(88), np.asarray(trains))\n",
    "plt.plot(np.arange(88), np.asarray(tests))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([1, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0,\n",
       "        0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 1, 0, 0,\n",
       "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]),\n",
       " array([0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0,\n",
       "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]))"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.asarray(Y_pred_tests)[:, 1598], Y_test[1598]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import f1_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.97133259316195519"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Y_pred_test_np = np.asarray(Y_pred_tests).T\n",
    "f1_score(Y_test.reshape(-1), Y_pred_test_np.reshape(-1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Среднее на обучении 0.99537139148\n",
      "Среднее на тесте 0.974584294649\n"
     ]
    }
   ],
   "source": [
    "train = []\n",
    "test = []\n",
    "for i in tqdm_notebook(range(88)):\n",
    "    train.append(roc_auc_score(Y_train[:, i],Y_pred_trains[i]))\n",
    "    test.append(roc_auc_score(Y_test[:, i],Y_pred_tests[i]))\n",
    "print (\"Среднее на обучении\", np.mean(train))\n",
    "print (\"Среднее на тесте\", np.mean(test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  },
  "widgets": {
   "state": {
    "3e4317383cc544fd95472de040f56433": {
     "views": [
      {
       "cell_index": 17
      }
     ]
    },
    "761cc0ec8baa456c9ca5102df963893d": {
     "views": [
      {
       "cell_index": 20
      }
     ]
    },
    "fa6b495d035e4d2db321fb3c39520712": {
     "views": [
      {
       "cell_index": 19
      }
     ]
    }
   },
   "version": "1.2.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}