Audio Coding Based on the Modulated Lapped Transform (MLT) and Set Partitioning in Hierarchical Trees

This paper presents an audio coder based on the combination of the Modulated Lapped Transform (MLT) with the Set Partitioning In Hierarchical Trees (SPIHT) algorithm. SPIHT allows scalable coding by transmitting more important information first in an efficient manner. The results presented reveal that the Modulated Lapped Transform (MLT) based scheme produces a high compression ratio for little or no loss of quality. A modification is introduced to SPIHT which further improves the performance of the algorithm when it is being used with uniform M-band transforms and masking. Further, the MLTSPIHT scheme is shown to achieve high quality synthesized audio at 54 kbps through subjective listening tests.