Scalable audio coding based on spatial perception in audio surveillance

A spatial perception based scalable audio coding in audio surveillance is presented in this paper. Sudden events in audio surveillance are often accompanied with high energy, fast changing energy or sound location. Current priority schemes of scalable audio coding are mainly based on the energy criterion or perception importance criterion, in which priorities are assigned according to high energy or low frequency which are more sensitive to auditory sensation. According to the energy and spatial cues in different frequency bands, this paper assigns the priorities to the sub-bands with fast changing location and energy of sound source in scalable audio coding. With zero extra side information and trivial added complexity, experimental results show that the perceptual quality is improved especially for the sound with high attention in audio surveillance.