MDCT domain parametric stereo audio coding

Parametric coding of multichannel audio has gained popularity for low bit rate audio coding applications such as digital audio broadcasting. Most of the existing algorithms use MDCT domain techniques for compressing the audio, while the spatialization parameter estimation is done in a different time-frequency domain. An MDCT domain parametric stereo coding algorithm which represents the stereo channels as the linear combination of the `sum´ channel derived from the stereo channels and a reverberated channel generated from the `sum´ channel has been reported in literature. Spatialization parameters are estimated at the encoder by taking the scaled sub-band projections of stereo channels on `sum´ and reverberated channel. This model is inadequate to represent the stereo image since only four parameters per sub-band are used as spatialization parameters. In this work we improve the perceptual quality of this MDCT domain parametric coder with an augmented parameter extraction scheme using an additional reverberated channel. Subjective evaluation using MUSHRA test illustrates that the new algorithm has increased the perceptual audio quality of the encoded audio signal significantly.