This should be straight forward to build in PyTorch or Keras. Not sure if there's a built in method, but here's a thread that discusses CNN-LSTM.