Effect of mp3 compression on sound localization
Kwan, Nicholas Alexis Shixian
Date of Issue2014
School of Electrical and Electronic Engineering
The majority of people today listen to music gathered online, through downloads or online web streaming services. Often, these audio files are highly compressed to produce a much smaller file size, for the ease of transmission through the Internet. The MP3 audio codec is a well-known and highly used lossy compression format. The MP3 encoding scheme is the use of a psychoacoustic model to determine redundant or irrelevant information fit for removal, such as frequencies beyond the human hearing range and sounds below masking thresholds. The bit rate used in the encoder determines the level of compression. The lowering of bit rate leads to a larger compression ratio, thus a smaller file size. Sound localization, one of the components of psychoacoustics, is the ability to pinpoint the source (real or virtual) of a sound heard in space. This space is commonly termed ‘soundstage’ among audiophiles. Sound localization depends on multiple aural cues based on perceptual phenomena such as pitch, loudness and duration. Considering how MP3 encoding employs psychoacoustics in removing information based on the bit rate set in the encoder, I derived my hypothesis as follows: Does compression rate affect accuracy of sound localization? To test if the accuracy of sound localization will decrease when compression rate increases, an experiment performed in this project measured participants’ (N=18) perceptual abilities to accurately localize sources when listening to a range of compressed audio in a stereo soundstage presented to them in an anechoic chamber. The results show that the position of sources was relevant (p-value = 1.22x10-15) and there was a relationship between compression rate and accuracy (p-value = 0.00398).
DRNTU::Visual arts and music
Final Year Project (FYP)
Nanyang Technological University