The SBC (subband coding) using the multirate filter bank was proposed in 1980, this coding method splits full audio spectrum into subbands, and assigns more bits to each subband with higher energy. If the power spectral density (PSD) of any subbands is approximately zero, then those subbands need not be assigned any bits at all. While reconstruction, the subband signals are synthesized to the original signal. The most important advantage of SBC is very high reduced data rate, but the acceptable signal quality. Nowadays, all of the SBC applied in audio signal coding use the Wavelet analysis. Its decimation ratios must be the power of 2. That is the shortcoming in the Wavelet analysis. This paper has proposed a new set of decimation ratios. When the SBC uses this set, it decreases data rate, compared to other used sets in condition of the same number of subbands.