Thu-3-2-9 Effect of Spectral Complexity Reduction and Number of Instruments on Musical Enjoyment with Cochlear Implants

Abstract: Although speech recognition technology for cochlear implants has continued to improve, music accessibility remains a challenge. Previous studies have shown that cochlear implant users may prefer listening to music that has been reengineered to be less complex. In this paper, we consider the combined effect of spectral complexity reduction and number of instruments playing on musical enjoyment with cochlear implants. Nine normal hearing listeners rated 200 10-second music samples on three enjoyment modalities (musicality, pleasantness, and naturalness) with and without the use of cochlear implant simulation. The music samples included 20 versions of the song ”Twinkle Twinkle Little Star” synthesized using one of five different instruments and with one to four instruments playing at once. The remaining 180 versions were created by reducing each sample’s spectral complexity to nine different levels using principal component analysis. The results showed a preference for less amounts of spectral complexity reduction for samples without cochlear implant simulation (P<.001), as well as a preference for fewer instruments for samples with cochlear implant simulation (P<.001). However, spectral complexity reduction was not a significant factor for samples with cochlear implant simulation, and a significant interaction effect between spectral complexity reduction and number of instruments was not found.