Objective and Subjective Evaluation of Loudspeaker Distortion Objektive und Subjektive Bewertung von Lautsprecherverzerrungen

A new auralization technique is presented for the objective and subjective assessment of loudspeakers in the large signal domain. On the basis of loudspeaker modeling and the linear, nonlinear and thermal parameters measured on a particular loudspeaker the instantaneous state variables (e.g. sound pressure output, displacement, temperature, ...) are calculated in real time for any input signal (test signal, music). The digital model allows a decomposition of the output signal into linear and nonlinear signal components and a novel kind of distortion analysis. This technique opens new possibilities for subjective listening test to assess the impact of each distortion component on sound quality and to optimize loudspeakers with respect to performance, size, weight and cost. Zusammenfassung Eine neue Auralizationstechnik wird zur objektiven and subjektiven Bewertung von Lautsprechern im Großsignalbereich vorgestellt. Mit Hilfe eines Lautsprechermodells und der am realen Lautsprecher gemessenen linearen, nichtlinearen und thermischen Parameter werden in einem digitalen Signalprozessor (DSP) die Zustandsvariablen (Schalldruck, Auslenkung, Temperatur, ...) für ein beliebige Signal (Testsignal, Musik) in Echtzeit berechnet. Mit Hilfe des digitalen Modells kann das abgestrahlte Schallsignal in lineare und nichtlineare Signalkomponenten zerlegt und eine Verzerrungsanalyse durchgeführt werden. Diese Technik eröffnet auch neue Möglichkeiten für subjektive Hörversuche, in denen der Einfluss der Verzerrungen auf die Wiedergabequalität untersucht und der Lautsprecher hinsichtlich Leistungsvermögen, Gewicht und Kosten optimiert werden kann. Introduction For many applications small, light-weight and cost-effective electro-acoustical transducers are required which produce the required sound pressure output at high efficiency and acceptable signal distortion. Thus, targeting for “loud”speakers the large signal performance becomes more and more important. The nonlinear and thermal mechanisms are subject of research for many years and substantial progress has been made in modeling, measurement and nonlinear control to compensate for loudspeaker distortion actively. The theory and available practical tools show the relationship between the physical causes (large signal parameters of the loudspeaker) and the nonlinear symptoms (distortion, amplitude compression, instabilities, ...) generated by the transducer for any input (special test or ordinary audio signals). The new auralization technique combines subjective and objective evaluation of loudspeaker systems and answers the following questions: Under which condition does signal distortion become audible ? How strong is impact on subjectively perceived sound quality ? What are the physical causes for the distortion ? Which role plays the stimulus and the listening conditions ? How can the performance-cost ratio be improved ? Is the loudspeaker optimal for the particular application ? These questions are important for defining product specification and to develop a loudspeaker meeting this targets. Electromechanical Transducer Audio signal Mechanoacoustical Transducer Sound Propagation Radiation Amplifier Crossover Room Interference p(r2) p(r1) p(r3) Sound Propagation Radiation Room Acoustics Sound Propagation Radiation Room Interference i(t) u(t) x(t) sound field Fig. 1. Signal flow in a sound reproduction system Loudspeaker Modeling in real time The Auralization techniques is based on the large signal model of the loudspeaker as illustrated in Fig. 1. The loudspeaker is a nonlinear and time-varying system having a single input (at the terminals) but multiple outputs (in the sound field). In the electro-mechanical transducer there are the dominant nonlinearities such as force factor Bl(x) of the electro-dynamical motor, the compliance C(x) of the suspension and the inductance L(x) of the voice coil depending on the voice coil displacement x as shown in Fig. In common woofers and other direct radiating transducers the cone vibration, radiation and propagation of the sound wave may be approximated by a linear system having a transfer function H(f,r) depending on the listening point r in the sound field. For a fixed listener position the reproduction is modeled by the upper branch in Fig. 2.