meetps / crnn-music-genre-classification

Tensorflow Implementation of Convolutional Recurrent Neural Networks for Music Genre Classification

crnn-music-genre-classification

Music Genre Classification using Convolutional Recurrent Neural Networks

Data

• Download GENRE data from Marsyas
• Extract the tarball in the directory which contains the cloned repo.

Running the code

CNN

• python cnn.py

CRNN

• python crnn.py

crnn-music-genre-classification's Issues

obtain a not ideal result

ask for help
why set to n_samples=1000 , my result auc gradually decrease,
and the loss is printed every batch_size, I find the loss not decreasing
my python version is 3.6, tensorflow version is 1.4.
and the code can run correctly,

read_data.py

labels = pd.read_csv(labels_file,header=0)
showing the error while reading the file.
Please help to rectify it.

crnn.py dynamic_rnn runtime error : ValueError

Hi

I have runtime error for execution crnn.py

My env :
OS : Windows 7
Python 3.5.3 Anaconda

Any solution to fix it ?

---

crnn.py line 67

gru1_out, state = tf.nn.dynamic_rnn (gru1, gru1_in, dtype=tf.float32, scope='gru1')

ValueError: Trying to share variable gru1/multi_rnn_cell/cell_0/gru_cell/gates/kernel,
but specified shape (64, 64) and found shape (160, 64).

crnn.py

on running crnn.py iam getting this error
I:\Coding\crnn-music-genre-classification-master\src>python crnn.py
C:\Users\dsemw\AppData\Local\Programs\Python\Python36\lib\site-packages\h5py_init_.py:36: FutureWarning: Conversion of the second argument of issubdtype from float to np.floating is deprecated. In future, it will be treated as np.float64 == np.dtype(float).type.
from ._conv import register_converters as _register_converters
Traceback (most recent call last):
File "crnn.py", line 94, in
X_test = rd.get_melspectrograms_indexed(test_indices)
File "I:\Coding\crnn-music-genre-classification-master\src\read_data.py", line 25, in get_melspectrograms_indexed
spectrograms = np.asarray([dp.log_scale_melspectrogram(i) for i in labels_dense['path'][index]])
File "I:\Coding\crnn-music-genre-classification-master\src\read_data.py", line 25, in
spectrograms = np.asarray([dp.log_scale_melspectrogram(i) for i in labels_dense['path'][index]])
File "I:\Coding\crnn-music-genre-classification-master\src\data_preprocess.py", line 21, in log_scale_melspectrogram
signal = signal[(n_sample-n_sample_fit)/2:(n_sample+n_sample_fit)/2]
TypeError: slice indices must be integers or None or have an index method

cnn.py prints AUC : 1.0 for all epoch ?

Hi

Thanks for your code.

I executed this code for cnn.py

All epoch print "AUC : 1.0" from start step to end step.

And,
for argmax mean, all epoch print "0.0"

Is this a good execution result ?

My Env is below :
Python 3.5.3
Anaconda 64 bit
Windows 7 64bit

how to make mp3 files into txt files?

how to make mp3 files into txt files?

error out of memory when n_samples=1000

W tensorflow/core/common_runtime/bfc_allocator.cc:275] Ran out of memory trying to allocate 6.59GiB. See logs for memory state.
W tensorflow/core/framework/op_kernel.cc:993] Resource exhausted: OOM when allocating tensor with shape[200,64,96,1440]
Traceback (most recent call last):
File "crnn.py", line 144, in
predictions = sess.run(predict_op, feed_dict=test_input_dict)
File "/home/ubuntu/anaconda3/envs/tf_1.0.1_py_2.7/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 767, in run
run_metadata_ptr)
File "/home/ubuntu/anaconda3/envs/tf_1.0.1_py_2.7/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 965, in _run
feed_dict_string, options, run_metadata)
File "/home/ubuntu/anaconda3/envs/tf_1.0.1_py_2.7/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1015, in _do_run
target_list, options, run_metadata)
File "/home/ubuntu/anaconda3/envs/tf_1.0.1_py_2.7/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1035, in _do_call
raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.ResourceExhaustedError: OOM when allocating tensor with shape[200,64,96,1440]
[[Node: Conv2D = Conv2D[T=DT_FLOAT, data_format="NHWC", padding="SAME", strides=[1, 1, 1, 1], use_cudnn_on_gpu=true, _device="/job:localhost/replica:0/task:0/gpu:0"](Reshape_1, Variable/read)]]
[[Node: Sigmoid/_87 = _Recvclient_terminated=false, recv_device="/job:localhost/replica:0/task:0/cpu:0", send_device="/job:localhost/replica:0/task:0/gpu:0", send_device_incarnation=1, tensor_name="edge_249_Sigmoid", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/cpu:0"]]

I can't train with the whole dataset, even 500 samples.