Abstract
The paper proposes an improved perceptual norm adjustment algorithm for ITU-T G.719, in which norm adjustment is based on segmental weighted Signal-to-Mask Ratio (SMR). This algorithm uses SMR to detect the location of harmonics, then weights SMR with the segmental spectral energy. SMR is computed from the reconstructed quantized MDCT coefficients. Finally the algorithm guides bit allocation with the adjusted norm and enhances psychoacoustic masking effect in bit allocation. Results of informal subjective listening tests show that the proposed method can improve the performance of G.719 slightly at low bit rates by using perceptual norm adjustment with segmental SMR.