In my thesis I explain auditory masking, and how it is exploited in lossy audio compression to produce smaller audio files, while the lossy audio sample remains indistinguishable from the lossless reference sample for the average person. I show how the Vorbis audio codec works, with an emphasis on audio encoding and decoding. I use Mathworks' MATLAB and Simulink to examine the process of decoding. With the help of the PEAQ model, I compare the quality of Vorbis' compression on different bitrates. I assembled a test, in which participants were asked to rate multiple lossy samples compared to a reference lossless sample. I implemented this subjective test in a JAVA program. I show the workings of this program, and the MySQL database i used to store the results. I analyzed the results using Mathworks' MATLAB. I look at the differences and similarities between the PEAQ model and my subjective test. Finally, I compare the ODG scores from the PEAQ model and my test’s results using Mathworks’ MATLAB.