The object of this project is to present a novel digital audio effect based on a real-time, windowed block-based FFT and inverse FFT. The effect is achieved by mirroring the spectrum, producing a sound effect ranging from a purer rendition of the original, through a rougher one, to a sound unrecognisable from the original. A mirror taking the shape of a conic section is constructed between certain partials, and the modified spectrum is created by reflecting the original spectrum in this mirror. The user can select the type and continuously vary the amount of curvature, typically ‘roughening’ the input sound quite gratifyingly. We demonstrate the system with live real-time audio via microphone.

Download

We present a novel method to adjust the perceived width of a phantom source by varying the deterministic inter channel time difference (ICT D) in a pair of signals over frequency. In contrast to given literature that focuses on random phase over frequency, our paper considers a deterministic approach that is open to a more systematic evaluation. Two allpass structures are described, finite impulse response (FIR) and infinite impulse response (IIR), for phase-based phantom source widening and evaluated in a formal listening test. Varying ICT D over frequency essentially alters the inter-aural cross correlation coefficient at the ears of a listener and in this way provides a robust way to control the auditory source width. The subjective evaluation results fully support our observations for both noise and speech signals.

Download

This article investigates the accuracy with which listeners can identify the spatial extent of distributed sound sources. Either the complementary frequency bands comprising a source signal or the individual grains of a granular synthesis-based stimulus were distributed directly on discrete loudspeakers. Loudspeakers were arranged either on the horizontal or the vertical axis. The algorithms were applied on white noise, an impulse train, and a rain drops stimulus. Absolute judgments of spatial extent were obtained separately for each orientation, algorithm, and stimulus using three different magnitudes of horizontal or vertical extent. Horizontal spatial extent judgments varied systematically with physical extent for all conditions in the experiment. The correspondence between perceived and actual vertical extent was poor. The time-based synthesis algorithm resulted in significantly larger judgments of spatial extent irrespective of orientation and stimulus compared to the frequency-based algorithm.

Download