In this paper

a speech subband coding system combined with time domain harmonic scaling algorithm has been studied.The filter banks used for splitting subband are designed by employing orthogonal and biothogonal wavelet transform

which are similar to discrimination property of human auditory.Moreover

we provide several quantization techniques at different bands.Our goal is to raise the compression ratio of speech greatly degree while preserves good objective decoding SNR and subjective qualities of reconstucted speech.