除了方文山，用TA你也能帮周杰伦写歌词了

周杰伦几乎陪伴了每个90后的青春，那如果AI写杰伦风格的歌词会写成怎样呢？

首先当然我们需要准备杰伦的歌词，这里一共收录了他的十几张专辑，近5000多行歌词。

原文档格式：

第一步数据预处理

def preprocess(data):
    """
    对文本中的字符进行替换，空格转换成逗号；换行变为句号。
    """
    data = data.replace(' ', '，')
    data = data.replace('\n', '。')
    words = jieba.lcut(data, cut_all=False) # 全模式切词

    return words

处理后结果：

前10个词: ['想要', '有', '直升机', '。', '想要', '和', '你', '飞到', '宇宙', '去']

将处理完的数据写入内存并将文本转换完数字

# 构造词典及映射
vocab = set(text)
vocab_to_int = {w: idx for idx, w in enumerate(vocab)}
int_to_vocab = {idx: w for idx, w in enumerate(vocab)}

# 转换文本为整数
int_text = [vocab_to_int[w] for w in text]

构建神经网络

a. 构建输入层

def get_inputs():
    inputs = tf.placeholder(tf.int32, [None, None], name='inputs')
    targets = tf.placeholder(tf.int32, [None, None], name='targets')
    learning_rate = tf.placeholder(tf.float32, name='learning_rate')
    return inputs, targets, learning_rate

b. 构建堆叠RNN单元

其中rnn_size指的是RNN隐层神经元个数

def get_init_cell(batch_size, rnn_size):
    lstm = tf.contrib.rnn.BasicLSTMCell(rnn_size)
    cell = tf.contrib.rnn.MultiRNNCell([lstm])

    initial_state = cell.zero_state(batch_size, tf.float32)
    initial_state = tf.identity(initial_state, 'initial_state')
    return cell, initial_state

c. Word Embedding

因为单词太多，所以需要进行embedding，模型中加入Embedding层来降低输入词的维度

def get_embed(input_data, vocab_size, embed_dim):
    embedding = tf.Variable(tf.random_uniform([vocab_size, embed_dim], -1, 1))
    embed = tf.nn.embedding_lookup(embedding, input_data)

    return embed

d. 构建神经网络，将RNN层与全连接层相连

其中cell为RNN单元； rnn_size: RNN隐层结点数量；input_data即input tensor；vocab_size：词汇表大小； embed_dim: 嵌入层大小

def build_nn(cell, rnn_size, input_data, vocab_size, embed_dim):
    embed = get_embed(input_data, vocab_size, embed_dim)
    outputs, final_state = build_rnn(cell, embed)

    logits = tf.contrib.layers.fully_connected(outputs, vocab_size, activation_fn=None)

    return logits, final_state

e. 构造batch

这里我们根据batch_size和seq_length分为len//(batch_size*seq_length)个batch，每个batch包含输入和对应的目标输出

def get_batches(int_text, batch_size, seq_length):
    '''
    构造batch
    '''
    batch = batch_size * seq_length
    n_batch = len(int_text) // batch

    int_text = np.array(int_text[:batch * n_batch]) # 保留能构成完整batch的数量

    int_text_targets = np.zeros_like(int_text)
    int_text_targets[:-1], int_text_targets[-1] = int_text[1:], int_text[0]

    # 切分
    x = np.split(int_text.reshape(batch_size, -1), n_batch, -1)
    y = np.split(int_text_targets.reshape(batch_size, -1), n_batch, -1)

    return np.stack((x, y), axis=1) # 组合

模型训练

from tensorflow.contrib import seq2seq
train_graph = tf.Graph()
with train_graph.as_default():
    vocab_size = len(int_to_vocab) # vocab_size
    input_text, targets, lr = get_inputs() # 输入tensor
    input_data_shape = tf.shape(input_text)
    # 初始化RNN
    cell, initial_state = get_init_cell(input_data_shape[0], rnn_size)
    logits, final_state = build_nn(cell, rnn_size, input_text, vocab_size, embed_dim)
    # 计算softmax层概率
    probs = tf.nn.softmax(logits, name='probs')
    # 损失函数
    cost = seq2seq.sequence_loss(
        logits,
        targets,
        tf.ones([input_data_shape[0], input_data_shape[1]]))
    # 优化函数
    optimizer = tf.train.AdamOptimizer(lr)
    # Gradient Clipping
    gradients = optimizer.compute_gradients(cost)
    capped_gradients = [(tf.clip_by_value(grad, -1., 1.), var) for grad, var in gradients if grad is not None]
    train_op = optimizer.apply_gradients(capped_gradients)

训练结果

Epoch  72 Batch   24/33   train_loss = 0.108
Epoch  75 Batch   25/33   train_loss = 0.104
Epoch  78 Batch   26/33   train_loss = 0.096
Epoch  81 Batch   27/33   train_loss = 0.111
Epoch  84 Batch   28/33   train_loss = 0.119
Epoch  87 Batch   29/33   train_loss = 0.130
Epoch  90 Batch   30/33   train_loss = 0.141
Epoch  93 Batch   31/33   train_loss = 0.138
Epoch  96 Batch   32/33   train_loss = 0.153
Model Trained and Saved

train_loss还不错，不过可能过拟合了。

最后让我们加载模型，看看生成情况

    # 加载模型
    loader = tf.train.import_meta_graph(save_dir + '.meta')
    loader.restore(sess, save_dir)
    # 获取训练的结果参数
    input_text, initial_state, final_state, probs = get_tensors(loaded_graph)
    # Sentences generation setup
    gen_sentences = [prime_word]
    prev_state = sess.run(initial_state, {input_text: np.array([[1]])})
    # 生成句子
    for n in range(gen_length):
        dyn_input = [[vocab_to_int[word] for word in gen_sentences[-seq_length:]]]
        dyn_seq_length = len(dyn_input[0])
        # 预测
        probabilities, prev_state = sess.run(
            [probs, final_state],
            {input_text: dyn_input, initial_state: prev_state})
        # 选择单词进行文本生成，用来以一定的概率生成下一个词
        pred_word = pick_word(probabilities[0][dyn_seq_length - 1], int_to_vocab)
        gen_sentences.append(pred_word)

哎哟不错哦！

最后的最后我还扩大了歌词库，这次引入了更多流行歌手，来看看效果吧。

好像更不错了！

如果你也喜欢杰伦，请点赞并分享生成的歌词。

点击关注，第一时间了解华为云新鲜技术~