Abstract

Musical instrument sounds have been shown to have distinct timbral and emotional characteristics, and when audio processes are applied to them, their timbral and emotional characteristics are changed. In this thesis, we investigated how audio processes change the perceived emotional characteristics of musical instrument sounds.

We first investigated the effects of MP3 compression on the emotional characteristics of instrument sounds, which has not been explored previously. Our results showed that MP3 compression strengthened neutral and negative emotional characteristics such as Scary and Sad, and weakened positive emotional characteristics such as Happy and Romantic. Interestingly, Angry was relatively unaffected by MP3 compression.

For artificial reverberation, since our previous research has shown that the distinctive emotional characteristics in musical instruments can be significantly changed with parametric reverberation, we would like to see whether the parametric reverberation results can be applied to real concert hall reverberation, namely convolution reverberation, as well. We would like to know whether these changes in character are relatively uniform or instrument-dependent as well.

Our finding shows that convolution reverberation had more pronounced effects on the emotional characteristics compared to parametric reverberation, yet there was a strong agreement in the results of parametric and convolution reverberations. For investigating into the underlying instrument space with reverberation, our results indicate that the underlying instrument space did not change much with both parametric and convolution reverberations, in terms of emotional characteristics. It means that reverberation time has a remarkably consistent effect on the emotional characteristics no matter whether parametric or convolution reverberation was used. It is also a reflection of their deep underlying functional similarities despite their fundamentally different implementations.

In terms of applications, our MP3 compression study will give listeners and music streaming service providers some preliminary benchmarks for understanding the emotional effects of MP3 compression on music. For the artificial reverberation studies, the relatively consistent rankings of emotional characteristics between the instruments certainly help each instrument retain its identity in different halls. Moreover, the instrument-independent behavior of concert halls is perhaps what helps distinguish a good music venue from a poor one. This can be an interesting avenue for future work.