speech-to-text/3.birnn-ctc-beam.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import librosa\n",
    "import os\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "from tqdm import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "wav_files = [f for f in os.listdir('./data') if f.endswith('.wav')]\n",
    "text_files = [f for f in os.listdir('./data') if f.endswith('.txt')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 2800/2800 [00:26<00:00, 107.58it/s]\n"
     ]
    }
   ],
   "source": [
    "inputs, targets = [], []\n",
    "for (wav_file, text_file) in tqdm(zip(wav_files, text_files), total = len(wav_files),ncols=80):\n",
    "    path = './data/' + wav_file\n",
    "    try:\n",
    "        y, sr = librosa.load(path, sr = None)\n",
    "    except:\n",
    "        continue\n",
    "    inputs.append(\n",
    "        librosa.feature.mfcc(\n",
    "            y = y, sr = sr, n_mfcc = 40, hop_length = int(0.05 * sr)\n",
    "        ).T\n",
    "    )\n",
    "    with open('./data/' + text_file) as f:\n",
    "        targets.append(f.read())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "inputs = tf.keras.preprocessing.sequence.pad_sequences(\n",
    "    inputs, dtype = 'float32', padding = 'post'\n",
    ")\n",
    "\n",
    "chars = list(set([c for target in targets for c in target]))\n",
    "num_classes = len(chars) + 2\n",
    "\n",
    "idx2char = {idx + 1: char for idx, char in enumerate(chars)}\n",
    "idx2char[0] = '<PAD>'\n",
    "char2idx = {char: idx for idx, char in idx2char.items()}\n",
    "\n",
    "targets = [[char2idx[c] for c in target] for target in targets]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pad_sentence_batch(sentence_batch, pad_int):\n",
    "    padded_seqs = []\n",
    "    seq_lens = []\n",
    "    max_sentence_len = max([len(sentence) for sentence in sentence_batch])\n",
    "    for sentence in sentence_batch:\n",
    "        padded_seqs.append(sentence + [pad_int] * (max_sentence_len - len(sentence)))\n",
    "        seq_lens.append(len(sentence))\n",
    "    return padded_seqs, seq_lens\n",
    "\n",
    "def sparse_tuple_from(sequences, dtype=np.int32):\n",
    "    indices = []\n",
    "    values = []\n",
    "\n",
    "    for n, seq in enumerate(sequences):\n",
    "        indices.extend(zip([n] * len(seq), range(len(seq))))\n",
    "        values.extend(seq)\n",
    "\n",
    "    indices = np.asarray(indices, dtype=np.int64)\n",
    "    values = np.asarray(values, dtype=dtype)\n",
    "    shape = np.asarray([len(sequences), np.asarray(indices).max(0)[1] + 1], dtype=np.int64)\n",
    "\n",
    "    return indices, values, shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pad_second_dim(x, desired_size):\n",
    "    padding = tf.tile([[0]], tf.stack([tf.shape(x)[0], desired_size - tf.shape(x)[1]], 0))\n",
    "    return tf.concat([x, padding], 1)\n",
    "\n",
    "class Model:\n",
    "    def __init__(\n",
    "        self,\n",
    "        num_layers,\n",
    "        size_layers,\n",
    "        learning_rate,\n",
    "        num_features,\n",
    "        dropout = 1.0,\n",
    "    ):\n",
    "        self.X = tf.placeholder(tf.float32, [None, None, num_features])\n",
    "        self.Y = tf.sparse_placeholder(tf.int32)\n",
    "        seq_lens = tf.count_nonzero(\n",
    "            tf.reduce_sum(self.X, -1), 1, dtype = tf.int32\n",
    "        )\n",
    "        self.label = tf.placeholder(tf.int32, [None, None])\n",
    "        self.Y_seq_len = tf.placeholder(tf.int32, [None])\n",
    "\n",
    "        def cells(size, reuse = False):\n",
    "            return tf.contrib.rnn.DropoutWrapper(\n",
    "                tf.nn.rnn_cell.LSTMCell(\n",
    "                    size,\n",
    "                    initializer = tf.orthogonal_initializer(),\n",
    "                    reuse = reuse,\n",
    "                ),\n",
    "                state_keep_prob = dropout,\n",
    "                output_keep_prob = dropout,\n",
    "            )\n",
    "\n",
    "        features = self.X\n",
    "        for n in range(num_layers):\n",
    "            (out_fw, out_bw), (\n",
    "                state_fw,\n",
    "                state_bw,\n",
    "            ) = tf.nn.bidirectional_dynamic_rnn(\n",
    "                cell_fw = cells(size_layers),\n",
    "                cell_bw = cells(size_layers),\n",
    "                inputs = features,\n",
    "                sequence_length = seq_lens,\n",
    "                dtype = tf.float32,\n",
    "                scope = 'bidirectional_rnn_%d' % (n),\n",
    "            )\n",
    "            features = tf.concat((out_fw, out_bw), 2)\n",
    "\n",
    "        logits = tf.layers.dense(features, num_classes)\n",
    "        time_major = tf.transpose(logits, [1, 0, 2])\n",
    "        decoded, log_prob = tf.nn.ctc_beam_search_decoder(time_major, seq_lens)\n",
    "        decoded = tf.to_int32(decoded[0])\n",
    "        self.preds = tf.sparse.to_dense(decoded)\n",
    "        self.cost = tf.reduce_mean(\n",
    "            tf.nn.ctc_loss(\n",
    "                self.Y,\n",
    "                time_major,\n",
    "                seq_lens\n",
    "            )\n",
    "        )\n",
    "        self.optimizer = tf.train.AdamOptimizer(\n",
    "            learning_rate = learning_rate\n",
    "        ).minimize(self.cost)\n",
    "        \n",
    "        preds = self.preds[:, :tf.reduce_max(self.Y_seq_len)]\n",
    "        masks = tf.sequence_mask(self.Y_seq_len, tf.reduce_max(self.Y_seq_len), dtype=tf.float32)\n",
    "        preds = pad_second_dim(preds, tf.reduce_max(self.Y_seq_len))\n",
    "        y_t = tf.cast(preds, tf.int32)\n",
    "        self.prediction = tf.boolean_mask(y_t, masks)\n",
    "        mask_label = tf.boolean_mask(self.label, masks)\n",
    "        self.mask_label = mask_label\n",
    "        correct_pred = tf.equal(self.prediction, mask_label)\n",
    "        correct_index = tf.cast(correct_pred, tf.float32)\n",
    "        self.accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/sparse_ops.py:1165: sparse_to_dense (from tensorflow.python.ops.sparse_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Create a `tf.sparse.SparseTensor` and use `tf.sparse.to_dense` instead.\n"
     ]
    }
   ],
   "source": [
    "tf.reset_default_graph()\n",
    "sess = tf.InteractiveSession()\n",
    "\n",
    "size_layers = 128\n",
    "learning_rate = 1e-3\n",
    "num_layers = 2\n",
    "batch_size = 32\n",
    "epoch = 50\n",
    "\n",
    "model = Model(num_layers, size_layers, learning_rate, inputs.shape[2])\n",
    "sess.run(tf.global_variables_initializer())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:56<00:00,  1.61it/s, accuracy=0.774, cost=13.9]  \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 1, average cost 30.895426, average accuracy 0.410282\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.66it/s, accuracy=0.77, cost=12.1] \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 2, average cost 13.515759, average accuracy 0.763336\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.67it/s, accuracy=0.774, cost=11.4]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 3, average cost 12.485374, average accuracy 0.766179\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.67it/s, accuracy=0.781, cost=10.9]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 4, average cost 11.965700, average accuracy 0.771188\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [01:04<00:00,  1.20it/s, accuracy=0.796, cost=10.6]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 5, average cost 11.654921, average accuracy 0.775891\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [01:01<00:00,  1.67it/s, accuracy=0.793, cost=10.4]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 6, average cost 11.420854, average accuracy 0.779490\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.73it/s, accuracy=0.796, cost=10.4]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 7, average cost 11.266212, average accuracy 0.783184\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:57<00:00,  1.72it/s, accuracy=0.785, cost=10.2]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 8, average cost 11.120960, average accuracy 0.784547\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:57<00:00,  1.71it/s, accuracy=0.804, cost=9.9] \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 9, average cost 10.962759, average accuracy 0.788159\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:57<00:00,  1.69it/s, accuracy=0.815, cost=9.66]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 10, average cost 10.801116, average accuracy 0.788978\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.68it/s, accuracy=0.807, cost=9.43]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 11, average cost 10.701082, average accuracy 0.789929\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:57<00:00,  1.72it/s, accuracy=0.815, cost=9.46]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 12, average cost 10.606546, average accuracy 0.790707\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.69it/s, accuracy=0.819, cost=9.22]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 13, average cost 10.441219, average accuracy 0.793228\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:57<00:00,  1.70it/s, accuracy=0.815, cost=8.95]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 14, average cost 10.284247, average accuracy 0.794920\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.70it/s, accuracy=0.822, cost=8.85]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 15, average cost 10.213211, average accuracy 0.796231\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.68it/s, accuracy=0.833, cost=8.49]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 16, average cost 10.213884, average accuracy 0.796875\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.68it/s, accuracy=0.822, cost=8.28]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 17, average cost 10.024258, average accuracy 0.798941\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.68it/s, accuracy=0.826, cost=8.03]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 18, average cost 9.865512, average accuracy 0.800763\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.66it/s, accuracy=0.841, cost=7.78]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 19, average cost 9.622712, average accuracy 0.804442\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.81it/s, accuracy=0.856, cost=7.1] \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 20, average cost 9.420062, average accuracy 0.808081\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.67it/s, accuracy=0.856, cost=7.03]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 21, average cost 9.190894, average accuracy 0.811815\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.57it/s, accuracy=0.885, cost=6.46]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 22, average cost 8.950453, average accuracy 0.815622\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.75it/s, accuracy=0.878, cost=6.34]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 23, average cost 8.695628, average accuracy 0.820073\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.72it/s, accuracy=0.874, cost=6.17]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 24, average cost 8.501432, average accuracy 0.822451\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.68it/s, accuracy=0.874, cost=6.06]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 25, average cost 8.280502, average accuracy 0.828445\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.71it/s, accuracy=0.889, cost=5.49]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 26, average cost 8.009401, average accuracy 0.833485\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:59<00:00,  1.71it/s, accuracy=0.893, cost=5.19]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 27, average cost 7.689209, average accuracy 0.839349\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.69it/s, accuracy=0.881, cost=5.15]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 28, average cost 7.454049, average accuracy 0.845384\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.68it/s, accuracy=0.881, cost=5.06]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 29, average cost 7.141874, average accuracy 0.851007\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.74it/s, accuracy=0.896, cost=4.79]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 30, average cost 6.954190, average accuracy 0.854842\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.65it/s, accuracy=0.926, cost=4.38]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 31, average cost 6.737239, average accuracy 0.859144\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.70it/s, accuracy=0.907, cost=4.45]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 32, average cost 6.555829, average accuracy 0.863140\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.70it/s, accuracy=0.933, cost=3.68]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 33, average cost 6.328900, average accuracy 0.866478\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.76it/s, accuracy=0.919, cost=3.34]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 34, average cost 6.238356, average accuracy 0.866374\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.66it/s, accuracy=0.948, cost=3.16]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 35, average cost 6.007752, average accuracy 0.871557\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.72it/s, accuracy=0.911, cost=3.29]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 36, average cost 5.648700, average accuracy 0.879011\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.67it/s, accuracy=0.952, cost=3.02]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 37, average cost 5.366171, average accuracy 0.886224\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.64it/s, accuracy=0.919, cost=3.13]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 38, average cost 5.212911, average accuracy 0.887460\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.70it/s, accuracy=0.937, cost=2.9] \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 39, average cost 5.055423, average accuracy 0.891845\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.65it/s, accuracy=0.907, cost=2.92]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 40, average cost 4.887426, average accuracy 0.895629\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:57<00:00,  1.77it/s, accuracy=0.963, cost=2.26]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 41, average cost 4.767081, average accuracy 0.897271\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.76it/s, accuracy=0.967, cost=2.38]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 42, average cost 4.469442, average accuracy 0.904982\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.77it/s, accuracy=0.937, cost=2.56]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 43, average cost 4.372556, average accuracy 0.904271\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.69it/s, accuracy=0.978, cost=2.04]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 44, average cost 4.227151, average accuracy 0.908447\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.70it/s, accuracy=0.967, cost=2.15]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 45, average cost 3.905823, average accuracy 0.914122\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:57<00:00,  1.71it/s, accuracy=0.978, cost=1.96]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 46, average cost 3.631991, average accuracy 0.919396\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.73it/s, accuracy=0.974, cost=1.76]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 47, average cost 3.457691, average accuracy 0.924996\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.67it/s, accuracy=0.967, cost=1.63]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 48, average cost 3.442457, average accuracy 0.924099\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:58<00:00,  1.78it/s, accuracy=0.985, cost=1.2] \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 49, average cost 3.342965, average accuracy 0.924908\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:57<00:00,  1.67it/s, accuracy=0.97, cost=1.28] "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 50, average cost 3.050437, average accuracy 0.930819\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "for e in range(epoch):\n",
    "    pbar = tqdm(\n",
    "        range(0, len(inputs), batch_size), desc = 'minibatch loop')\n",
    "    total_cost, total_accuracy = 0, 0\n",
    "    for i in pbar:\n",
    "        batch_x = inputs[i : min(i + batch_size, len(inputs))]\n",
    "        y = targets[i : min(i + batch_size, len(inputs))]\n",
    "        batch_y = sparse_tuple_from(y)\n",
    "        batch_label, batch_len = pad_sentence_batch(y, 0)\n",
    "        _, cost, accuracy = sess.run(\n",
    "            [model.optimizer, model.cost, model.accuracy],\n",
    "            feed_dict = {model.X: batch_x, model.Y: batch_y, \n",
    "                         model.label: batch_label, model.Y_seq_len: batch_len},\n",
    "        )\n",
    "        total_cost += cost\n",
    "        total_accuracy += accuracy\n",
    "        pbar.set_postfix(cost = cost, accuracy = accuracy)\n",
    "    total_cost /= (len(inputs) / batch_size)\n",
    "    total_accuracy /= (len(inputs) / batch_size)\n",
    "    print('epoch %d, average cost %f, average accuracy %f'%(e + 1, total_cost, total_accuracy))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "real: say the word cheek\n",
      "predicted: say the word chek\n"
     ]
    }
   ],
   "source": [
    "import random\n",
    "\n",
    "random_index = random.randint(0, len(targets) - 1)\n",
    "batch_x = inputs[random_index : random_index + 1]\n",
    "print(\n",
    "    'real:',\n",
    "    ''.join(\n",
    "        [idx2char[no] for no in targets[random_index : random_index + 1][0]]\n",
    "    ),\n",
    ")\n",
    "batch_y = sparse_tuple_from(targets[random_index : random_index + 1])\n",
    "pred = sess.run(model.preds, feed_dict = {model.X: batch_x})[0]\n",
    "print('predicted:', ''.join([idx2char[no] for no in pred]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}