Wednesday February 17th 3.00pm
Computer Science Auditorium (CSG-01)
Alex Wilson is currently a PhD student at the University of Salford’s Acoustics Research Centre, investigating the perception of quality in sound recordings, focussing on the production of popular music. This research focuses on the mechanics and psychoacoustics of audio engineering, specifically the task of mix-engineering, addressing three fundamental questions:
· What does mix-engineering involve?
· What makes a good mix?
· How can good mixes be generated automatically?
He obtained a B.Sc in Experimental Physics from NUI Maynooth in 2008 and a B.Eng in Audio Technology from University of Salford in 2013. During this time, twelve months were spent as an R&D engineer at Sennheiser GmbH, in Hannover, Germany, and, more recently, six months as a visiting researcher at the Centre for Digital Music at Queen Mary University of London. He maintains interests in digital audio processing, psychoacoustics and the art of record production.
To further the development of intelligent music production tools, towards generating mixes that would realistically be created by a human mix-engineer, and enjoyed by a listener, it is important to understand what kind of mixes are typically created by human mix-engineers. What can be achieved by mixing? How different can mixes be from one another? How much do different mix-engineers differ from one another?
1501 audio samples were gathered, representing the alternate mixes of 10 songs. We have investigated the distribution of low level audio signal features, over mixes of each song and the entire dataset. Quantitative analysis reveals that mix-engineers mostly vary the perceived “loudness”, “brightness” and “width” of the mix. Typical ranges for these dimensions have been determined. By plotting the full set of mixes in a low-dimensional space, it is observed that it is possible to mix any of the songs considered to have the general loudness/brightness/width characteristics of another.
This novel research can impact a number of theoretical and practical problems. In addition to intelligent audio production, the work has implications for audio education and for the field of music information retrieval, towards improved algorithms for tasks such as tempo estimation, artist identification and genre classification.