A voice signal encoding and decoding method, device, and codec system are provided. The coding method includes: encoding an input voice signal to obtain a broadband code stream, where the broadband code stream includes a core layer bit stream and an extension enhancement layer bit stream (101); compressing the core layer bit stream to obtain a compressed code stream (102); and packing the compressed code stream and the extension enhancement layer bit stream to obtain a packed code stream (103). The core layer bit stream is compressed, and the compressed code stream and the extension enhancement layer bit stream are packed, thereby reducing transmission bandwidth occupied by the input voice signal. Since the broadband voice encoding is performed on the input voice signal, a broadband voice code stream is transmitted by using narrowband transmission bandwidth, thereby improving the cost performance of voice signal transmission.