A method for upmixing of single-channel audio signals for stereophonic sound reproduction in real-time is presented. To this end, the input signal is decomposed into a foreground signal and a background signal using frequency domain processing. The background signal is decorrelated using a network of nested allpass filters. The intensity of the decorrelation is controlled using a computational model for the perceived intensity of decor-relation. The foreground sound sources like singers and soloists are reproduced in the center of the stereo image. The proposed method enables upmixing from mono to stereo signals (and can also be applied to enhance the perceived width of a stereo recording) with low latency, moderate computational load and low memory requirements. It produces output signals with a high sound quality and is suitable for automotive and low-bitrate streaming applications.
