Concepts for Hybrid Audio Coding Schemes Based on Parametric Techniques
Bernd Edler, Heiko Purnhagen · 1998
Recently an audio coding technique based on a signal representation by parameters describing harmonic tones, individual sinusoidal components and noise components has been presented, which shows its highest efficiency at very low bit rates. In this paper concepts for combination with other techniques for extending the applicability at higher bit rates and for improving the quality in the presence of speech are presented. 1 Introduction A recently presented object based analysis/synthesis audio coder for very low bit rates is based on a signal representation which describes different components of an audio signal by appropriate parameter sets [1]. Together with transform coding and speech coding techniques it has been integrated in the MPEG-4 Audio development [2]. For harmonic tones the most important parameters are the fundamental frequency and the amplitudes of the partials. Individual sinusoidal components are represented by their frequencies and amplitudes, while noise components ...