With Tensorflow/Keras, is there a way to split up input data for training so I don’t run out of RAM?

Question

I am using tensorflow with Keras to process audio data through conv1d layers, however, I am running out RAM (only 8GB available). I have an input .wav and a target .wav for