Issue link: http://read.uberflip.com/i/324330
2014 Loudspeaker Industry Sourcebook 21 INDUSTRY REPORTS Figure 1: The mean preference ratings of six popular over-the-ear headphones are ranked based on ratings from trained listeners. The cascading graphs show the headphone's left and right frequency response with the average perceived spectral balance indicated by the dotted green line. measurements need to accurately capture this effect if they are to accurately predict how they sound. Headphone Target Response The current recommended target-response curves for headphones are diffuse-field (DF) or free-field (FF). Many headphone manufacturers and researchers are beginning to question the recommendation on the basis that it makes headphones sound too thin and bright (see Resources). Most stereo recordings are optimized to sound best through loudspeakers in rooms that produce neither DF nor FF responses, but something in between. A better approach may be to design a headphone with a target response that simulates the in-room response of an accurate loudspeaker calibrated in a reference listening room. To test this idea, trained listeners rated the sound quality of two different headphones equalized to various target-response curves based on three different DF curves, a FF curve, and two new target responses based on a loudspeaker in a room (see Resources). The unequalized headphones were also included as a hidden reference. Figure 2 plots the preference ratings of the different headphone target responses averaged across all listeners and programs. The two new target responses (Target 1 and Target 2) based on an accurate loudspeaker's in-room response were significantly preferred over both the DF and the FF options, and the original unequalized headphone (the Audeze LCD2). The experimental evidence suggests that a headphone's sound can be significantly improved when it simulates an accurate loudspeaker's in-room response. Headphone Sound influences The popularity of Beats By Dre headphones, particularly among young audio enthusiasts, has led many marketing departments to believe that headphone sound preferences may vary with age, listening experience, or even nationality. While these factors have, so far, not been found to influence decisions on loudspeaker sound or audio data compression, the hypothesis has never been tested for headphones (see Resources). To t h at e n d, a vir t u al h e a dp h o n e t e s t methodology was developed to enable different headphones to be measured and then virtualized through a calibrated replicator headphone (see Resources). In this way, the tests are truly double blind since potential tactile and other biases (e.g., brand, price, fashion, and celebrity endorsement) are removed from the test. First, a validation study was conducted to determine if the virtual method produced similar subjective ratings to the standard method. The correlation between methods was sufficiently accurate (r = 0.85) to draw valid conclusions. Then, double-blind listening tests were conducted during which more than 200 listeners from Canada, Germany, China, and the US gave preference ratings for four different virtualized headphones. The headphones included two audiophile models (the Sennheiser Electronic HD800 and the Audeze LCD2), the popular Beats by Dre Studio Edition, and the replicator headphone equalized to the average preferred in-room response of a loudspeaker calibrated in the HARMAN reference listening room. Figure 2: The mean preference ratings of different headphone target-response curves are simulated by applying EQ to the Audeze LCD2 headphone. Target 1 and Target 2 are based on the measured in-room response of an accurate loudspeaker in the HARMAN reference listening room. The No EQ category signifies the Audeze headphone with no EQ applied.