• A hybrid virtual bass system for optimized steady-state and transient performance

      Hill, Adam J.; Hawksford, Malcolm O. J.; University of Essex (IEEE, 2010-09)
      Bandwidth extension of a constrained loudspeaker system is regularly achieved employing nonlinear bass synthesis. The method operates on the doctrine of the missing fundamental whereby humans infer the presence of a fundamental tone when presented with a signal consisting of higher harmonics of said tone. Nonlinear devices and phase vocoders are commonly used for signal generation; both exhibiting deficiencies. A system is proposed where the two approaches are used in tandem via a mixing algorithm to suppress these deficiencies. Mixing is performed by signal transient content analysis in the frequency domain using constant-Q transforms. The hybrid approach is rated subjectively against various nonlinear device and phase vocoder techniques using the MUSHRA test method.
    • Wide-area psychoacoustic correction for problematic room-modes using nonlinear bass synthesis

      Hill, Adam J.; Hawksford, Malcolm O. J.; University of Essex (Audio Engineering Society, 2011-11)
      Small room acoustics are characterized by a limited number of dominant low-frequency room-modes that result in wide spatio-pressure variations that traditional room correction systems may find elusive to correct over a broad listening area. A psychoacoustic-based methodology is proposed whereby signal components coincident only with problematic modes are filtered and substituted by virtual bass components to forge an illusion of the suppressed frequencies. Although this approach can constitute a standalone correction system, the impetus for development is for use within well-established correction methodologies. A scalable and hierarchical approach is studied using subjective evaluation to confirm uniform wide-area performance. Bass synthesis exploits parallel nonlinear and phase vocoder generators with outputs blended as a function of transient and steady-state signal content.