Sound quality evaluation of high-speed train interior noise by adaptive Moore loudness algorithm

Abstract

An online experiment to acquire the interior noise of a China Railways High-speed (CRH) train showed that it was mainly composed of middle-low frequency components and could not be described properly by linear or A-weighted sound pressure level (SPL). Thus, the appropriate way to evaluate the high-speed train interior noise is to use sound quality parameters, and the most important is loudness. To overcome the disadvantages of the existing loudness algorithms, a novel signal-adaptive Moore loudness algorithm (AMLA) based on the equivalent rectangular bandwidth (ERB) spectrum was introduced. The validation reveals that AMLA can obtain higher accuracy and efficiency, and the simulated dark red noise conforms best to the high-speed train interior noise by loudness and auditory assessment. The main loudness component of the interior noise is below 27.6 ERB rate (erbr), and the sound quality of the interior noise is relatively stable between 300–350 km/h. The specific loudness components among 12–15 erbr stay invariable throughout the acceleration or deceleration process while components among 20–27 erbr are evidently speed related. The unusual random noise is effectively identified, which indicates that AMLA is an appropriate method for sound quality assessment of the high-speed train under both steady and transient conditions.