speech-to-text/7.wavenet.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 [00:44<00:00, 62.68it/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_causal(x, size, rate):\n",
    "    pad_len = (size - 1) * rate\n",
    "    return tf.pad(x, [[0, 0], [pad_len, 0], [0, 0]])\n",
    "\n",
    "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",
    "        num_blocks = 3,\n",
    "        block_size = 128,\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 residual_block(x, size, rate, block):\n",
    "            with tf.variable_scope('block_%d_%d' % (block, rate), reuse = False):\n",
    "                conv_filter = tf.layers.conv1d(\n",
    "                    x,\n",
    "                    x.shape[2] // 4,\n",
    "                    kernel_size = size,\n",
    "                    strides = 1,\n",
    "                    padding = 'same',\n",
    "                    dilation_rate = rate,\n",
    "                    activation = tf.nn.tanh,\n",
    "                )\n",
    "                conv_gate = tf.layers.conv1d(\n",
    "                    x,\n",
    "                    x.shape[2] // 4,\n",
    "                    kernel_size = size,\n",
    "                    strides = 1,\n",
    "                    padding = 'same',\n",
    "                    dilation_rate = rate,\n",
    "                    activation = tf.nn.sigmoid,\n",
    "                )\n",
    "                out = tf.multiply(conv_filter, conv_gate)\n",
    "                out = tf.layers.conv1d(\n",
    "                    out,\n",
    "                    block_size,\n",
    "                    kernel_size = 1,\n",
    "                    strides = 1,\n",
    "                    padding = 'same',\n",
    "                    activation = tf.nn.tanh,\n",
    "                )\n",
    "                return tf.add(x, out), out\n",
    "        forward = tf.layers.conv1d(self.X, block_size, kernel_size = 1, strides = 1, padding = 'SAME')\n",
    "        zeros = tf.zeros_like(forward)\n",
    "        for i in range(num_blocks):\n",
    "            for r in [1, 2, 4, 8, 16]:\n",
    "                forward, s = residual_block(forward, size=7, rate=r, block=i)\n",
    "                zeros = tf.add(zeros,s)\n",
    "        forward = tf.layers.conv1d(zeros, block_size, kernel_size = 1, strides = 1, padding = 'SAME')\n",
    "        logits = tf.layers.conv1d(zeros, num_classes, kernel_size = 1, strides = 1, padding = 'SAME')\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-4\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 [00:42<00:00,  2.30it/s, accuracy=0.607, cost=18.7] \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 1, average cost 31.322521, average accuracy 0.383164\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:42<00:00,  2.42it/s, accuracy=0.726, cost=14.8]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 2, average cost 17.311649, average accuracy 0.687664\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:42<00:00,  2.31it/s, accuracy=0.73, cost=13.8] \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 3, average cost 15.412181, average accuracy 0.739036\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:42<00:00,  2.36it/s, accuracy=0.793, cost=12.7]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 4, average cost 14.523385, average accuracy 0.754283\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:43<00:00,  2.36it/s, accuracy=0.781, cost=12]  \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 5, average cost 14.059295, average accuracy 0.759557\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.46it/s, accuracy=0.752, cost=11.7]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 6, average cost 13.705538, average accuracy 0.763334\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:42<00:00,  2.53it/s, accuracy=0.778, cost=11.9]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 7, average cost 13.493875, average accuracy 0.767596\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:42<00:00,  2.23it/s, accuracy=0.778, cost=11.7]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 8, average cost 13.325244, average accuracy 0.770417\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.57it/s, accuracy=0.789, cost=11.7]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 9, average cost 13.132063, average accuracy 0.771028\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.34it/s, accuracy=0.793, cost=11.6]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 10, average cost 13.050689, average accuracy 0.771950\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.45it/s, accuracy=0.789, cost=11.4]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 11, average cost 12.943646, average accuracy 0.773550\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.45it/s, accuracy=0.785, cost=11.5]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 12, average cost 12.848162, average accuracy 0.775142\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.36it/s, accuracy=0.781, cost=11.4]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 13, average cost 12.763524, average accuracy 0.776144\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.56it/s, accuracy=0.793, cost=11.6]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 14, average cost 12.721658, average accuracy 0.776208\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.31it/s, accuracy=0.785, cost=11.3]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 15, average cost 12.630298, average accuracy 0.777150\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.50it/s, accuracy=0.793, cost=10.9]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 16, average cost 12.611158, average accuracy 0.777942\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.53it/s, accuracy=0.789, cost=10.9]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 17, average cost 12.521884, average accuracy 0.778383\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.42it/s, accuracy=0.793, cost=11.1]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 18, average cost 12.477354, average accuracy 0.777964\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.57it/s, accuracy=0.789, cost=11.1]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 19, average cost 12.451472, average accuracy 0.777937\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.36it/s, accuracy=0.793, cost=10.7]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 20, average cost 12.375108, average accuracy 0.779235\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.34it/s, accuracy=0.785, cost=10.8]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 21, average cost 12.340024, average accuracy 0.779885\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.66it/s, accuracy=0.789, cost=10.6]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 22, average cost 12.288662, average accuracy 0.779781\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.48it/s, accuracy=0.793, cost=10.8]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 23, average cost 12.275069, average accuracy 0.780598\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.39it/s, accuracy=0.789, cost=11.1]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 24, average cost 12.256266, average accuracy 0.779737\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.29it/s, accuracy=0.789, cost=10.8]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 25, average cost 12.211515, average accuracy 0.781889\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.50it/s, accuracy=0.789, cost=10.8]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 26, average cost 12.169458, average accuracy 0.780740\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.57it/s, accuracy=0.789, cost=10.6]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 27, average cost 12.189863, average accuracy 0.781492\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.37it/s, accuracy=0.789, cost=10.6]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 28, average cost 12.078650, average accuracy 0.780658\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.33it/s, accuracy=0.793, cost=10.7]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 29, average cost 12.053977, average accuracy 0.782076\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.57it/s, accuracy=0.789, cost=10.6]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 30, average cost 12.023519, average accuracy 0.780967\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.20it/s, accuracy=0.796, cost=10.3]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 31, average cost 11.997221, average accuracy 0.782247\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.57it/s, accuracy=0.785, cost=10.4]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 32, average cost 11.983206, average accuracy 0.781721\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.39it/s, accuracy=0.789, cost=10.3]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 33, average cost 11.940457, average accuracy 0.781590\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:41<00:00,  2.68it/s, accuracy=0.793, cost=10.3]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 34, average cost 11.949307, average accuracy 0.781798\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.64it/s, accuracy=0.793, cost=10.2]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 35, average cost 11.899714, average accuracy 0.782660\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.50it/s, accuracy=0.789, cost=10.3]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 36, average cost 11.898101, average accuracy 0.783081\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.55it/s, accuracy=0.785, cost=9.92]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 37, average cost 11.869951, average accuracy 0.782689\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.66it/s, accuracy=0.785, cost=10.3]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 38, average cost 11.840547, average accuracy 0.782597\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.58it/s, accuracy=0.793, cost=10.1]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 39, average cost 11.807605, average accuracy 0.783186\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.40it/s, accuracy=0.785, cost=9.98]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 40, average cost 11.778183, average accuracy 0.783243\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:38<00:00,  2.55it/s, accuracy=0.789, cost=9.95]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 41, average cost 11.742411, average accuracy 0.783374\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.50it/s, accuracy=0.785, cost=9.63]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 42, average cost 11.670368, average accuracy 0.785182\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.42it/s, accuracy=0.789, cost=9.77]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 43, average cost 11.698776, average accuracy 0.784518\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:38<00:00,  2.70it/s, accuracy=0.796, cost=9.72]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 44, average cost 11.638369, average accuracy 0.784811\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:40<00:00,  2.65it/s, accuracy=0.793, cost=9.74]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 45, average cost 11.621436, average accuracy 0.784622\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.51it/s, accuracy=0.789, cost=9.92]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 46, average cost 11.585062, average accuracy 0.785294\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.48it/s, accuracy=0.789, cost=9.73]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 47, average cost 11.570401, average accuracy 0.784454\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.48it/s, accuracy=0.793, cost=9.5] \n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 48, average cost 11.531226, average accuracy 0.785691\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.38it/s, accuracy=0.789, cost=9.72]\n",
      "minibatch loop:   0%|          | 0/88 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 49, average cost 11.534660, average accuracy 0.784504\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "minibatch loop: 100%|██████████| 88/88 [00:39<00:00,  2.39it/s, accuracy=0.789, cost=9.46]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 50, average cost 11.519573, average accuracy 0.784982\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 five\n",
      "predicted: say the word oae\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]))"
   ]
  }
 ],
 "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
}