keras ctc loss error: InvalidArgumentError: Saw a non-null label following a null label

在用keras-tf backend做语音识别网络时，采用MFCC特征值输入，LSTM网络和ctc loss function，出现了以下错误：

2018-07-02 11:32:45.861523: W tensorflow/core/framework/op_kernel.cc:1318] OP_REQUIRES failed at ctc_loss_op.cc:166 : Invalid argument: Saw a non-null label (index >= num_classes- 1) following a null label, batch: 4 num_classes: 29 labels:
Traceback (most recent call last):
  File "/media/chutz/000206BE0003636E/ASRT_SpeechRecognition/train_mspeech.py", line 47, in <module>
    ms.TrainModel(datapath, epoch = 50, batch_size = 8, save_step = 1000, filename= modelpath)
  File "/media/chutz/000206BE0003636E/ASRT_SpeechRecognition/SpeechModel25.py", line 215, in TrainModel
    #self._model.fit_generator(yielddatas, save_step, nb_worker=2)
  File "/home/chutz/anaconda3/lib/python3.5/site-packages/keras/legacy/interfaces.py", line 91, in wrapper
    return func(*args, **kwargs)
  File "/home/chutz/anaconda3/lib/python3.5/site-packages/keras/engine/training.py", line 2230, in fit_generator
    class_weight=class_weight)
  File "/home/chutz/anaconda3/lib/python3.5/site-packages/keras/engine/training.py", line 1883, in train_on_batch
    outputs = self.train_function(ins)
  File "/home/chutz/anaconda3/lib/python3.5/site-packages/keras/backend/tensorflow_backend.py", line 2482, in __call__
    **self.session_kwargs)
  File "/home/chutz/anaconda3/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 900, in run
    run_metadata_ptr)
  File "/home/chutz/anaconda3/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 1135, in _run
    feed_dict_tensor, options, run_metadata)
  File "/home/chutz/anaconda3/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 1316, in _do_run
    run_metadata)
  File "/home/chutz/anaconda3/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 1335, in _do_call
    raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.InvalidArgumentError: Saw a non-null label (index >= num_classes - 1) following a null label, batch: 4 num_classes: 29 labels:
         [[Node: ctc/CTCLoss = CTCLoss[_class=["loc:@training/SGD/gradients/ctc/CTCLoss_grad/mul"], ctc_merge_repeated=true, ignore_longer_outputs_than_inputs=false, preprocess_collapse_repeated=false, _device="/job:localhost/replica:0/task:0/device:CPU:0"](ctc/Log/_123, ctc/ToInt64/_125, ctc/GatherNd, ctc/Squeeze_1/_127)]]
         [[Node: training/SGD/gradients/ctc/CTCLoss_grad/