This paper presents an audio compression method based on wavelet sub-band quantization and coding, and proposes a coder based on that method. The proposed coder uses the wavelet packet transform in order to obtain the critical bands of the human auditory system. Some results of the MPEG psychoacoustic model 2 are used in the coding of wavelet coefficients. The MPEG results are transformed to the wavelet domain in order to determinate the quantizer type and the number of quantization levels for each wavelet subband. The transform method is also proposed. The coder uses scalar and vector quantization methods according with the sensibility of the human auditory system for each wavelet subband. The entropy coding is also used in order to improve the performance of the proposed coder. The results of the subjective evaluation demonstrate that the proposed coder achieves transparent coding of the monophonic CD signals at bit rates of 80-96 Kbit/s. © 2005 IEEE.