Cultural background, level of exposure, and social pressures are only a handful of the factors that impact which albums we listen to on our iPods every day (Hennig et al., 2011). Musical preference seems to be something decided by countless reasons, however,  researchers have recently discovered that this is not the case, and that there exist certain universal properties of music. Key amongst these is flicker noise (deviations from the true rhythm) seen in musical pitch and loudness fluctuations over long time scales (i.e. the length of a musical composition), which itself can be  represented by a spectral power density of the 1/ƒ type (where ƒ represents frequency; Voss and Clarke, 1978).

The 1/ƒ noise regime depicts long-range correlated (LRC) fluctuations, and it is hypothesized that this 1/ƒ noise is characteristic of complex systems. 1/ƒ noise is encountered by many scientific disciplines in physical systems where rhythm plays a prominent role, such as wrist motion, coordination, heartbeat and stride series, mental rotation, simple reaction time, and of course musical performances (Gilden, Thornton, and Mallon, 1995; Diniz et al., 2011; Hennig et al., 2011). However, the LRC fluctuations seen in each are not identical, indicating the 1/ƒ noise regime is not entirely universal. The fluctuations can be represented by 1/ƒ^^β, where β is a scaling exponent used to determine the persistence, or correlation, of deviations in the system. Weaker fluctuations correspond to systems where β < 1, while highly correlated series are seen when β > 1. The flicker noise most commonly observed in human musical performances has a scaling exponent of  β ≈ 1, and this is termed ‘super’ persistant (Hennig et al., 2011).

By studying rhythmic tasks involving the hands, feet, and voice as performed by musicians, researchers have revealed that LRC fluctuations are present in all musical rhythms, whether performed by an amateur or professional. The fluctuations as represented by 1/ƒ^^β differed from one musician to the next as each performed differently, leading to a relatively wide range of β values. Nonetheless, the 1/ƒ noise regime was seen in each rhythm (Hennig et al., 2011). The observation of varying scaling exponents  is a result of the fact that regardless of a musician’s skill level they will always play differently and imperfectly. Often, individuals anticipate the coming beat in a composition, leading to a slightly inaccurate rhythm that eventually deviates from the ideal or ‘perfect’ pattern. This is because LRC fluctuations are a generic feature of musical rhythms, meaning that a small rhythmic flux does not only alter the beat for a short time – the beat is altered even after tens of seconds (Hennig et al., 2011).

The most interesting aspect of these findings is not that musicians are innately imperfect. Instead, it is that we, as listeners of musical performances, greatly prefer rhythmic fluctuations with flicker noise (Hennig et al., 2011). It is common practice in today’s music industry to generate compositions using precise computers (Figure 1), and then ‘humanize’ the sound by applying white noise fluctuations to the musical sequence. Editing the sequence involves using audio software to decompose the beats, reorder them according to deviations in a random time series, and then recombine the track (Hennig et al., 2011). However, this method does not take into account the flicker noise inherent in human performances, and as such when we listen to similar musical sequences, one being performed by a human and one being computer generated, we show a strong preference to the human version. Furthermore, we also perceive the human performance to be more precise, even though the exact opposte is true (Hennig et al., 2011).

These results open the possibility of developing novel humanizing methods for use in the music industry (Hennig et al., 2010), hthey also reveal that we preferentially enjoy human musical performances. Even though the music we choose to listen to is a result of many different factors, we each share one unifying characteristic in musical taste.  When considering that this universal characteristic is a result of flicker noise, which happens to have the seemingly ever present 1/ƒ noise regime, a coincidence can be seen, warranting a great deal of further study.

 

References

Diniz A., et al., 2011. Contemporary theories of 1/ƒ noise in motor control. Human Movement Science, 30(5), p.889-905.

Gilden D., Thornton T., and Mallon M., 1995. 1/ƒ noise in human cognition. Science, 267(5205), p.1837-1839.

Hennig H., et al., 2011. The Nature and Perception of Fluctuations in Human Musical Rhythms. PLoS ONE, 6(10), doi: 10.1371/journal.pone.0026457.

Hennig H., Fleischmann R., Theis F., and Geisel T, 2010. Method and device for humanizing musical sequences. US Patent No. 7,777,123.

Voss, R. and Clarke, J., 1978. “1/ƒ noise” in music: Music from 1/ƒ noise. Journal of the Acoustical Society of America, 63(1), p.258-263.

 

Images Cited

[Sound Board] 2009. [image online] Available at: <http://themixtapemonster.wordpress.com/2009/09/08/5-best-almost-free-tools-for-music-production-enthusiasts/> [Accessed 19 November 2011].