the program crashes for short audio clips (less than 4 seconds). Any thoughts what could be wrong? is there any requirements on the input length?
W tensorflow/core/framework/op_kernel.cc:1202] OP_REQUIRES failed at conv_ops.cc:384 : Invalid argument: computed output size would be negative [[Node: conv1d_8/convolution/Conv2D = Conv2D[T=DT_FLOAT, data_format="NHWC", dilations=[1, 1, 1, 1], padding="VALID", strides=[1, 1, 2, 1], use_cudnn_on_gpu=true, _device="/job:localhost/replica:0/task:0/device:CPU:0"](conv1d_8/convolution/ExpandDims, conv1d_8/convolution/ExpandDims_1)]]

when loading model:

Traceback (most recent call last):
  File "soundnet.py", line 104, in <module>
    prediction = predict_scene_from_audio_file('railroad_audio.wav')
  File "soundnet.py", line 88, in predict_scene_from_audio_file
    model = build_model()
  File "soundnet.py", line 26, in build_model
    model_weights = np.load('models/sound8.npy').item()

How can we use this code to predict object categories?
The original code
http://soundnet.csail.mit.edu/
can predict object and scene at the same time.

Thanks!

At first, thanks for wonderful work from author, I appreciate.
I got a lot of bugs when I import keras follow authors. These bugs take me 2 hours.
We aren't supposed to install Keras separately when you use tf2. Use the Keras bundled with the Tensorflow you have installed.
Finally, I solved it.
We can change soundnet.py to this:

import numpy as np
import librosa
from tensorflow.keras.layers import BatchNormalization, Activation, Conv1D, MaxPooling1D, ZeroPadding1D, InputLayer
from tensorflow.keras.models import Sequential

Hi,
I tried loading the pretrained model in sound8.npy and got this error. What could be the problem? python version? tensorflow version?

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/zahra/anaconda3/envs/tensorflow/lib/python3.6/site-packages/numpy/lib/npyio.py", line 431, in load
    "Failed to interpret file %s as a pickle" % repr(file))
OSError: Failed to interpret file 'sound8.npy' as a pickle