(n-gram level)Generate Lyrics from all Ed Sheeran songs.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Generate Lyrics from all Ed sheeran songs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.preprocessing.sequence import pad_sequences\n",
    "from keras.layers import Embedding, LSTM, Dense\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.callbacks import EarlyStopping\n",
    "from keras.models import Sequential\n",
    "import keras.utils as ku "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>artist</th>\n",
       "      <th>song</th>\n",
       "      <th>link</th>\n",
       "      <th>text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ABBA</td>\n",
       "      <td>Ahe's My Kind Of Girl</td>\n",
       "      <td>/a/abba/ahes+my+kind+of+girl_20598417.html</td>\n",
       "      <td>Look at her face, it's a wonderful face  \\nAnd...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ABBA</td>\n",
       "      <td>Andante, Andante</td>\n",
       "      <td>/a/abba/andante+andante_20002708.html</td>\n",
       "      <td>Take it easy with me, please  \\nTouch me gentl...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ABBA</td>\n",
       "      <td>As Good As New</td>\n",
       "      <td>/a/abba/as+good+as+new_20003033.html</td>\n",
       "      <td>I'll never know why I had to go  \\nWhy I had t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ABBA</td>\n",
       "      <td>Bang</td>\n",
       "      <td>/a/abba/bang_20598415.html</td>\n",
       "      <td>Making somebody happy is a question of give an...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ABBA</td>\n",
       "      <td>Bang-A-Boomerang</td>\n",
       "      <td>/a/abba/bang+a+boomerang_20002668.html</td>\n",
       "      <td>Making somebody happy is a question of give an...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  artist                   song                                        link  \\\n",
       "0   ABBA  Ahe's My Kind Of Girl  /a/abba/ahes+my+kind+of+girl_20598417.html   \n",
       "1   ABBA       Andante, Andante       /a/abba/andante+andante_20002708.html   \n",
       "2   ABBA         As Good As New        /a/abba/as+good+as+new_20003033.html   \n",
       "3   ABBA                   Bang                  /a/abba/bang_20598415.html   \n",
       "4   ABBA       Bang-A-Boomerang      /a/abba/bang+a+boomerang_20002668.html   \n",
       "\n",
       "                                                text  \n",
       "0  Look at her face, it's a wonderful face  \\nAnd...  \n",
       "1  Take it easy with me, please  \\nTouch me gentl...  \n",
       "2  I'll never know why I had to go  \\nWhy I had t...  \n",
       "3  Making somebody happy is a question of give an...  \n",
       "4  Making somebody happy is a question of give an...  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd \n",
    "data = pd.read_csv('songdata.csv')\n",
    "data.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data[(data['artist']=='Ed Sheeran')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"It's alright to cry even my dad does sometimes  \\nSo don't wipe your eyes  \\nTears remind you you're alive  \\nIt's alright to die cause death the only thing you haven't tried  \\nBut just for tonight hold on  \\nSo live life like you're giving all  \\nCause you act like you are  \\nGo ahead and just live it up  \\nGo on and tell me your path  \\n  \\nIt's alright to shake  \\nEven my hand does sometimes  \\nSo inside the rage Against the dying of the light  \\nIt's alright to say that death's  \\nThe only thing you haven't tried  \\nBut just for today hold on  \\nSo live life like you're giving all  \\nCause you act like you are  \\nGo ahead and just live it up  \\nGo on and tell me your path  \\n  \\nGo ahead and just live it up  \\nGo on and tell me your path and hold on\\n\\n\""
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "l = list(data['text'])\n",
    "l[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'its alright to cry even my dad does sometimes  so dont wipe your eyes  tears remind you youre alive  its alright to die cause death the only thing you havent tried  but just for tonight hold on  so live life like youre giving all  cause you act like you are  go ahead and just live it up  go on and tell me your path    its alright to shake  even my hand does sometimes  so inside the rage against the dying of the light  its alright to say that deaths  the only thing you havent tried  but just for today hold on  so live life like youre giving all  cause you act like you are  go ahead and just live it up  go on and tell me your path    go ahead and just live it up  go on and tell me your path and hold on'"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import string\n",
    "def clean_text(txt):\n",
    "    txt = \"\".join(v for v in txt if v not in string.punctuation).lower()\n",
    "    txt = txt.replace(\"\\n\",\"\")\n",
    "    return txt \n",
    "\n",
    "corpus = [clean_text(x) for x in l]\n",
    "corpus[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "53"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(corpus)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "tokenizer = Tokenizer()\n",
    "\n",
    "def get_sequence_of_tokens(corpus):\n",
    "    ## tokenization\n",
    "    tokenizer.fit_on_texts(corpus)\n",
    "    total_words = len(tokenizer.word_index) + 1\n",
    "    \n",
    "    ## convert data to sequence of tokens \n",
    "    input_sequences = []\n",
    "    for line in corpus:\n",
    "        token_list = tokenizer.texts_to_sequences([line])[0]\n",
    "        for i in range(1, len(token_list)):\n",
    "            n_gram_sequence = token_list[:i+1]\n",
    "            input_sequences.append(n_gram_sequence)\n",
    "    return input_sequences, total_words\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[29, 482],\n",
       " [29, 482, 7],\n",
       " [29, 482, 7, 270],\n",
       " [29, 482, 7, 270, 137],\n",
       " [29, 482, 7, 270, 137, 5],\n",
       " [29, 482, 7, 270, 137, 5, 361],\n",
       " [29, 482, 7, 270, 137, 5, 361, 747],\n",
       " [29, 482, 7, 270, 137, 5, 361, 747, 426],\n",
       " [29, 482, 7, 270, 137, 5, 361, 747, 426, 32],\n",
       " [29, 482, 7, 270, 137, 5, 361, 747, 426, 32, 37]]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "inp_sequences, total_words = get_sequence_of_tokens(corpus)\n",
    "inp_sequences[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def generate_padded_sequences(input_sequences):\n",
    "    max_sequence_len = max([len(x) for x in input_sequences])\n",
    "    input_sequences = np.array(pad_sequences(input_sequences, maxlen=max_sequence_len, padding='pre'))\n",
    "    \n",
    "    predictors, label = input_sequences[:,:-1],input_sequences[:,-1]\n",
    "    label = ku.to_categorical(label, num_classes=total_words)\n",
    "    return predictors, label, max_sequence_len\n",
    "\n",
    "predictors, label, max_sequence_len = generate_padded_sequences(inp_sequences)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def create_model(max_sequence_len, total_words):\n",
    "    input_len = max_sequence_len - 1\n",
    "    model = Sequential()\n",
    "    \n",
    "    # Add Input Embedding Layer\n",
    "    model.add(Embedding(total_words, 10, input_length=input_len))\n",
    "    \n",
    "    # Add Hidden Layer 1 - LSTM Layer\n",
    "    model.add(LSTM(100))\n",
    "    \n",
    "    # Add Output Layer\n",
    "    model.add(Dense(total_words, activation='softmax'))\n",
    "\n",
    "    model.compile(loss='categorical_crossentropy', optimizer='adam')\n",
    "    \n",
    "    return model\n",
    "\n",
    "model = create_model(max_sequence_len, total_words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      " - 378s - loss: 6.0563\n",
      "Epoch 2/10\n",
      " - 367s - loss: 5.7208\n",
      "Epoch 3/10\n",
      " - 364s - loss: 5.5771\n",
      "Epoch 4/10\n",
      " - 400s - loss: 5.3932\n",
      "Epoch 5/10\n",
      " - 447s - loss: 5.2287\n",
      "Epoch 6/10\n",
      " - 445s - loss: 5.0589\n",
      "Epoch 7/10\n",
      " - 391s - loss: 4.8913\n",
      "Epoch 8/10\n",
      " - 416s - loss: 4.7293\n",
      "Epoch 9/10\n",
      " - 467s - loss: 4.5667\n",
      "Epoch 10/10\n",
      " - 334s - loss: 4.4060\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x118a93252e8>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(predictors, label, epochs=10, verbose=2)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def generate_text(seed_text, next_words, model, max_sequence_len):\n",
    "    for _ in range(next_words):\n",
    "        token_list = tokenizer.texts_to_sequences([seed_text])[0]\n",
    "        token_list = pad_sequences([token_list], maxlen=max_sequence_len-1, padding='pre')\n",
    "        predicted = model.predict_classes(token_list, verbose=0)\n",
    "        \n",
    "        output_word = \"\"\n",
    "        for word,index in tokenizer.word_index.items():\n",
    "            if index == predicted:\n",
    "                output_word = word\n",
    "                break\n",
    "        seed_text += \" \"+output_word\n",
    "    return seed_text.title()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Look At The Homeless Life For Me\n"
     ]
    }
   ],
   "source": [
    "print (generate_text(\"look at\", 5, model, max_sequence_len))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "I Am With The Only One\n"
     ]
    }
   ],
   "source": [
    "print (generate_text(\"i am\", 4, model, max_sequence_len))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "My Life And The Same One And A Cruel Eight I Love Her I Know My Mind\n"
     ]
    }
   ],
   "source": [
    "print (generate_text(\"my life\", 15, model, max_sequence_len))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pretty decent results"
   ]
  }
 ],
 "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}