A method, medium, and system scalably encoding/decoding audio/speech. The method includes splitting an input signal into a low frequency band signal that is lower than a predetermined frequency and a high frequency band signal that is higher than the predetermined frequency, scalably encoding the split low frequency band signal into a core layer and one or more extension layers and then decoding the encoded core layer and the encoded extension layers, generating an error signal by using the split low frequency band signal and a decoded signal of the encoded core layer and the encoded extension layers, and encoding the error signal and the high frequency band signal into a signal-to-noise ratio (SNR) enhancement layer and a bandwidth extension layer.