Open Mddct opened 1 year ago
I have the same issue when trying to do a transer learning example of tensorflow which strongly depends on tensorflow-io: https://www.tensorflow.org/tutorials/audio/transfer_learning_audio
I need tensorflow-io becaue it is easier to work with it when using tensorflow datasets. So it would be great if a workaround would be mentioned or the bug will be fixed.
I have replaced the tensorflow-io audio processing code with librosa on mac but it is still not very handy to work with variable sized arrays when not having tensorflow datasets or tensors.
So hopefully this is fixed soon.
python3 env: 3.9.6 tfio version: 0.34.0