search

zzw922cn / Automatic_Speech_Recognition

End-to-end Automatic Speech Recognition for Madarian and English in Tensorflow
MIT License
2.84k stars 537 forks source link

Problem with DeepSpeech2 #79

Open fanlamda opened 6 years ago

fanlamda commented 6 years ago

I have an error when running with the DeepSpeech2 model. Any idea with this problem is welcome. I have tried tensorflow version 1.1-1.4. All failed Traceback (most recent call last): File "libri_train.py", line 265, in runner.run() File "libri_train.py", line 152, in run model = model_fn(args, maxTimeSteps) File "build/bdist.linux-x86_64/egg/speechvalley/models/deepSpeech2.py", line 105, in init File "build/bdist.linux-x86_64/egg/speechvalley/utils/utils.py", line 25, in wrapper File "build/bdist.linux-x86_64/egg/speechvalley/models/deepSpeech2.py", line 134, in build_graph File "build/bdist.linux-x86_64/egg/speechvalley/models/deepSpeech2.py", line 44, in build_deepSpeech2 File "/home/fanlei1/anaconda2/lib/python2.7/site-packages/tensorflow/python/layers/normalization.py", line 586, in batch_normalization return layer.apply(inputs, training=training) File "/home/fanl/anaconda2/lib/python2.7/site-packages/tensorflow/python/layers/base.py", line 671, in apply return self.call(inputs, *args, *kwargs) File "/home/fanl/anaconda2/lib/python2.7/site-packages/tensorflow/python/layers/base.py", line 575, in call outputs = self.call(inputs, args, **kwargs) File "/home/fanl/anaconda2/lib/python2.7/site-packages/tensorflow/python/layers/normalization.py", line 395, in call return self._fused_batch_norm(inputs, training=training) File "/home/fanl/anaconda2/lib/python2.7/site-packages/tensorflow/python/layers/normalization.py", line 302, in _fused_batch_norm training, _fused_batch_norm_training, _fused_batch_norm_inference) File "/home/fanl/anaconda2/lib/python2.7/site-packages/tensorflow/python/layers/utils.py", line 203, in smart_cond pred_value = constant_value(pred) File "/home/fanl/anaconda2/lib/python2.7/site-packages/tensorflow/python/layers/utils.py", line 233, in constant_value raise TypeError('pred must be a Tensor, a Variable, or a Python bool.') TypeError: pred must be a Tensor, a Variable, or a Python bool.

michaelklachko commented 6 years ago

DeepSpeech2 model appears to be broken. Your particular error just means that the argument isTraining is not supplied (because it does not exist). But even if you fix this error, there will be plenty more...