Learning spectro-temporal representations of complex sounds with parameterized neural networks | Read Paper on Bytez