R-TTT module with modified residual signal for improving multichannel audio signal accuracy

Spatial audio coding is a technique that capable of representing multichannel audio signals as a lower number of audio channels accompanied by spatial parameters and residual signal which will be useful for recreating the original multi-channel audio signals. Moving Picture Expert Group (MPEG) Surround, an international standard developed based on spatial audio coding, specifies Reverse Two-To-Three (R-TTT) module to extend stereo audio, consisted of left and right channels, into three audio channels: left, centre, and right channels based on Channel Prediction Coefficient (CPC) as spatial parameter and residual signal. In this paper, a modified residual signal is proposed to provide a better audio waveform reconstruction in the decoder side by minimising distortion caused by quantisation of CPC. Our experiments show that the waveform accuracy in terms of Signal-to-Noise Ratio (SNR) gets improved as high as 11 dB while the subjective test shows that the proposed method does not reduce perceptual quality, in terms of Subjective Difference Grade (SDG) score, of the reconstructed audio signals.