search

BYRTIMO / END-TO-END-SPEECH-ENHANCEMENT-BASED-ON-DISCRETE-COSINE-TRANSFORM

18 stars 4 forks source link

这句报错，还有noisy是指的clean和noise合成之后音频吧 #1

Open wbjnpu opened 4 years ago

wbjnpu commented 4 years ago

grads = opt.compute_gradients(loss_fn, var_list=[var for var in tf.trainable_variables()])

BYRTIMO commented 4 years ago

对的，错误我看下

wbjnpu commented 4 years ago

对的，错误我看下

Traceback (most recent call last): File "train.py", line 122, in train(cfg) File "train.py", line 58, in train grads = opt.compute_gradients(loss_fn, var_list=[var for var in tf.trainable_variables()]) File "/home/npu/anaconda3/envs/wbj/lib/python3.6/site-packages/tensorflow/python/training/optimizer.py", line 519, in compute_gradients colocate_gradients_with_ops=colocate_gradients_with_ops) File "/home/npu/anaconda3/envs/wbj/lib/python3.6/site-packages/tensorflow/python/ops/gradients_impl.py", line 630, in gradients gate_gradients, aggregation_method, stop_gradients) File "/home/npu/anaconda3/envs/wbj/lib/python3.6/site-packages/tensorflow/python/ops/gradients_impl.py", line 814, in _GradientsHelper lambda: grad_fn(op, out_grads)) File "/home/npu/anaconda3/envs/wbj/lib/python3.6/site-packages/tensorflow/python/ops/gradients_impl.py", line 408, in _MaybeCompile return grad_fn() # Exit early File "/home/npu/anaconda3/envs/wbj/lib/python3.6/site-packages/tensorflow/python/ops/gradients_impl.py", line 814, in lambda: grad_fn(op, out_grads)) File "/home/npu/anaconda3/envs/wbj/lib/python3.6/site-packages/tensorflow/python/ops/math_grad.py", line 328, in _UnsortedSegmentSumGrad return _GatherDropNegatives(grad, op.inputs[1])[0], None, None File "/home/npu/anaconda3/envs/wbj/lib/python3.6/site-packages/tensorflow/python/ops/math_grad.py", line 296, in GatherDropNegatives for in range(gathered.shape.ndims - is_positive.shape.ndims): TypeError: unsupported operand type(s) for -: 'NoneType' and 'int' 这个，哪里的数据类型不一样吧

BYRTIMO commented 4 years ago

嗯嗯，应该是最后一句“gathered.shape.ndims - is_positive.shape.ndims”中减号两边数据类型不同，参照“https://stackoverflow.com/questions/47100024/unsupported-operand-types-for-nonetype-and-int”，你可以看下。

wbjnpu commented 4 years ago

嗯嗯，应该是最后一句“gathered.shape.ndims - is_positive.shape.ndims”中减号两边数据类型不同，参照“https://stackoverflow.com/questions/47100024/unsupported-operand-types-for-nonetype-and-int”，你可以看下。

这修改tensorflow源码么

BYRTIMO commented 4 years ago

我没有修改过，我的tensorflow版本为1.13

wbjnpu commented 4 years ago

请问有预训练模型提供一下么

Huer-H commented 4 years ago

你好，问下这个noisy路径下的文件和clean路径下的文件是一一对应的吗？ 比如： mix/a.wav clean/a.wav？ mix和clean路径下的数量是一致的吗？

Huer-H commented 4 years ago

是不是按照徐勇老师那个代码合成的noisy呢？ 多个mix对应一个clean？