Stereo to mono down-mix is a key component of parametric stereo coding to drastically reduce the bit rate, but at the same time it is also an irreversible process that is a potential source of undesirable artifacts. This paper aims to reduce typical distortions induced by down-mixing, such as signal cancellation, comb filtering or unnatural instabilities. Two down-mixing methods are designed with different trade-offs between natural timbre and energy preservation based on simple rules that ensure low complexity. The results of a listening test show that both the proposed methods have a substantial advantage over the passive down-mix, while being very competitive compared to more computationally demanding active down-mixing approaches. The proposed methods are, therefore, particularly well suited to low complexity stereo coding schemes, such as those required for communication applications.
Download Now (1.5 MB)