Reporting on a talk by Pauline Larrouy-Maestri from the Max Plank Institute for Empirical Aesthetics in Germany, with an academic background in psychology, speech therapy, music, and pedagogy, 12th November 2015
Did Marilyn actually sing in tune? Sounds simple, but according to Pauline Larrouy-Maestri there are many factors that contribute to our perception of whether a singer is ‘in tune’ or ‘out of tune’. In a talk given to the Music Mind and Brain class, she explained that although we can perceive very small deviations in mistuning, this does not always hamper our enjoyment of a performance. Even operatic singers don’t always sing in tune. When Larrouy-Maestri, Magis and Morsomme (2014a) objectively measured it. They showed that their singing voice can deviate up to 100 cents from a target tone, which is way more than we would usually find acceptable in a different type of singer; but somehow it can still sound ok to the listener! As bizarre as this sounds, this is because the fundamental frequency is only one factor amongst other aspects such as shimmer, jitter, vibrato, tempo or timbre (Larrouy-Maestri, Magis and Morsomme, 2014b) in their singing performance.
Focusing on performances of occasional singers (i.e. without specific vocal training), Pauline Larrouy-Maestri, Lévêque, Schön, Giovanni, and Morsomme (2013) started a series of experiments to see what aspects of singing lead listeners to hear incorrect intonation. In one of their earliest experiments in 2013, 166 sung performances of happy birthday were analyzed in terms of error using computer softwares (Larrouy-Maestri & Morsomme, 2014) and then by eighteen judges with musical expertise. They then compared the judges ratings to errors identified objectively by the computer software, and confirmed that deviation from target intervals, was subjectively considered as an error by the listeners. More recently, they replicated this study with laymen, i.e., judges without any formal training in music and found a similar pattern of results (Larrouy-Maestri, Magis, Grabenhorst, & Morsomme, 2015). These studies confirm that the more an interval is deviated, the more the performance will be perceived as out of tune. But if this is so, where does out of tune singing end and in tune singing begin?
In order to find out the tolerance of listeners when listening to melodies, Pauline Larrouy-Maestri kept busy and conducted another series of experiments. First she recorded simple melodies and manipulated them in order to create perfectly in tune performances. Then, deviations (enlargement or compression) were integrated to the melodies to obtain several versions of the same meoldy, from in tune to clearly out of tune. Finally, people were asked to listen to the melodic sequences and to determine whether the sequence was in tune or out of tune. The errors were decreased in size to find each person’s limit in their ability to detect a tuning error. The main conclusion is that non-musicians can readily identify tiny differences in pitch, equivalent to just a quartertone, and sometimes even smaller! On top of this they found that musicians were able to pick up on tuning errors of one tenth of a semitone. Errors like these were picked up regardless of whether the interval manipulated was ascending or descending, and regardless of where it appeared in the melody or on with type of interval the error occurred. So the auditory system can in fact hear tiny tuning errors, which means people should be able to hear Marilyn as singing out of tune even if this performance does not contain “big” errors! Perhaps familiarity with a song influences your perception of it being sung ‘out of tune’ because you would be more likely to pick up on tuning errors? For instance, if your favorite Lady GaGa song was being sung out of tune, you might be more likely to notice and not enjoy it. But when this theory was put to the test, Larrouy-Maestri was not able to find any evidence to support it. In fact she said that perhaps you may be familiar with an out of tune version, which definitely won’t help matters!
Interestingly, singers rarely start a tone on the exact pitch. In collaboration with Peter Pfordresher, at the Auditory Perception and Action Lab in Buffalo (US), Larrouy-Maestri observed that singers usually start either a semitone above or below where they want to be and then they glide toward the correct note, leading to a fluctuation of tone known as a ‘scoop’, (which can be visualized as sliding towards the specific point you need to reach). The poor-pitch singers recorded in Pfordresher & Mantell (2013) do this too, except their scoops are almost double the size, leading to a lot more fluctuation before they sing the correct note. This is because it takes inaccurate singers longer to achieve the correct motor adjustment in their vocal cords to produce the correct sound.
Looking at this further, with the presentation of four different types of melodies to 102 people – including tuning errors in the starting scoop, the end scoop, or the part in the middle- they observed that listeners are sensitive to pitch deviations of the stable part of the tone (as expected) but also at the scoops. The results support the combination of both averaging and sequential processes when listening to melodies.
Overall the talk was very informative with a lot of individual experiments outlined, providing a wealth of evidence in relation to sung pitch perception. However orientation from an overarching theory or perceptual model would help to understand the connections and relationships between the many pieces of evidence she presented.
Now, listen to Marilyn again and see if you can hear the scoops or tuning errors. But does this hamper your enjoyment of the song? In effect, the question becomes not, does the singer sing in tune, but does it matter? Perhaps beauty truly lies in the ears of the beholder!
Claire Howlin & Michelle Ulor
Larrouy-Maestri, P., Magis, D., Grabenhorst, M., & Morsomme, D. (2015). Layman versus Professional Musician: Who Makes the Better Judge? PLoS ONE, 10(8), e0135394. doi:10.1371/journal.pone.0135394
Larrouy-Maestri, P., Magis, D., & Morsomme, D. (2014a). The evaluation of vocal accuracy: The case of operatic singing voices. Music Perception, 32(1), 1-10. DOI: 10.1525/MP.2014.32.1.1
Larrouy-Maestri, P., Magis, D., & Morsomme, D. (2014b). Effects of melody and technique on acoustical and musical features of western operatic singing voices. Journal of Voice, 28(3), 332-340. doi:10.1016/j.jvoice.2013.10.019
Larrouy-Maestri, P., & Morsomme, D. (2014). Criteria and tools for objectively analysing the vocal accuracy of a popular song. Logopedics Phoniatrics Vocology, 39, 11-18. DOI: 10.3109/14015439.2012.696139
Larrouy-Maestri, P., Lévêque, Y., Schön, D., Giovanni, A., & Morsomme, D. (2013). The evaluation of singing voice accuracy: A comparison between subjective and objective methods. Journal of Voice, 27(2), 259e1-e5. DOI: 10.1016/j.jvoice.2012.11.003
Pfordresher, P. Q., Mantell, J. T. (2013). Singing with yourself: Evidence for an inverse modelling account for poor-pitch singing. Cognitive Psychology, 70, 31-57.