Authors:Deruty, Emmanuel; Pachet, François; Roy, Pierre
Affiliation:Sony Computer Science Laboratory, Paris, France
Sophisticated production and mixing techniques in the rock music genre were developed in the second half of the 20th century, but there is no unified theory of mixing tracks. This paper explores the relationship between loudness and spectrum of individual tracks using an ad hoc 3-dimensional representation of spectra as applied to 55 rock multitracks. Given good monitoring conditions, i.e. at higher monitoring levels and on full-range monitors, comparatively brighter tracks are mixed softer, and comparatively darker tracks are mixed louder. Track loudness appears to be a linear function of its spectral centroid, and audio engineers appear to be concerned with the perceived spectral balance. On the other hand, given difficult monitoring conditions, i.e. at lower monitoring levels and on budget monitors, track spectrum and loudness are set conjointly so that each track is optimally understandable.
Download: PDF (HIGH Res) (1.8MB)
Download: PDF (LOW Res) (799KB)
Authors:Lee, Chung; Horner, Andrew
Affiliation:The Information Systems Technology and Design Pillar, Singapore University of Technology and Design, Singapore; Department of Computer Science and Engineering, Hong Kong University of Science and Technology, Kowloon, Hong Kong
Electronic instruments and music synthesizers have been criticized as being unnatural because the notes are uniform without natural variations. When a musician plays a sequence of “identical” notes, each one is in fact somewhat different from the others. In this study, eight sustained musical instrument sounds were randomly altered by a time-invariant process to determine the degree to which spectral alterations could be detected with repeated notes. Listeners were asked to discriminate each randomly altered repeated note sequence from an original unaltered sequence. The results showed that spectrally altered repeated note sequences were significantly more discriminable than single tones of the same duration. Correlation analysis confirmed that spectral incoherence correlated significantly for three alteration factors in the single-note stimuli and two alteration factors in the repeated note sequences.
Download: PDF (HIGH Res) (1.1MB)
Download: PDF (LOW Res) (259KB)
Authors:Wu, Bin; Horner, Andrew; Lee, Chung
Affiliation:Department of Computer Science and Engineering, Hong Kong University of Science and Technology, Kowloon, Hong Kong; The Information Systems Technology and Design Pillar, Singapore University of Technology and Design, Singapore
While melody, rhythm, and harmony are important emotional triggers in music, there has been little consideration of timbre. The authors designed a series of listening tests to compare the emotionality of sounds from eight wind and bowed stringed instruments. The violin, trumpet, and clarinet were best at evoking the emotions of happy, joyful, heroic, and comic. Conversely, the horn and flute evoked the emotions of sad and depressed. The oboe was emotionally neutral. Emotions correlated with average spectral centroid and spectral centroid deviation. The results suggest that the even/odd harmonic ratio is perhaps the most salient timbral feature after attack time and brightness. This research has direct implications for musicians and audio engineers who are doing orchestration for such applications as computer games, film sound, and stage music.
Download: PDF (HIGH Res) (733KB)
Download: PDF (LOW Res) (351KB)
Authors:Iversen, Niels Elkjær; Knott, Arnold
Affiliation:Technical University of Denmark, Kgs. Lyngby, Denmark
In order to properly evaluate a loudspeaker amplifier, it is useful to test its performance under realistic conditions. This implies a load that emulates the complex impedance of an actual loudspeaker driver with its interacting mechanical resonances. A loudspeaker emulator is therefore a useful tool to replace a fixed resistor. Results from an experimental emulation validate the utility for small signals at low frequencies. The design allows the resonant frequencies to be changed by tuning two resistors. Even though the emulation does not exactly match an actual driver, suggestions are provided to improve the approximations.
Download: PDF (HIGH Res) (2.4MB)
Download: PDF (LOW Res) (457KB)
World experts in forensic audio analysis gathered in London to share the latest research and practice in the field. Among the topics covered were “roomprints,” speaker identification from VoIP communications, the analysis of quantizing levels, and the handling of gunshot recordings. The distortion evident in mains power signals can provide a good clue to geographical location, and forensic musicology needs to become more scientific.
Download: PDF (635KB)
Affiliation:ROLI, London, UK
Predicting the future is like predicting the weather: even small currents will move a forecast completely off course. Much larger currents are at play today: throughout the financial world; within debates about alternative energy and climate change; in the increasingly dysfunctional sphere of international patent law; across the face of the Middle East and the developing world. Where science, craft, and the most abstract of the arts intersect, audio engineers are uniquely placed to respond to adversity. We handle complexity, solve technical problems adulterated by political dimensions, and respond smartly to imperfect information. Whether a retailer, recording engineer, product designer, or educator, our job demands optimism, subtlety, and diplomacy. The world has never needed us more.
Download: PDF (414KB)