Blending Between Bassoon and Horn Players: an Analysis of Timbral Adjustments During Musical Performance
Total Page:16
File Type:pdf, Size:1020Kb
144 Sven-Amin Lembke, Scott Levine, & Stephen McAdams BLENDING BETWEEN BASSOON AND HORN PLAYERS: AN ANALYSIS OF TIMBRAL ADJUSTMENTS DURING MUSICAL PERFORMANCE SVEN-AMIN LEMBKE,SCOTT LEVINE,& MONG THE MANY AIMS OF ORCHESTRATION, STEPHEN MCADAMS the combination of instruments into a blended McGill University, Montreal, Canada A timbre is one that is most relevant perceptually. Although decisions concerning orchestration can be ACHIEVING A BLENDED TIMBRE BETWEEN TWO primarily guided by personal preference, blend relies instruments is a common aim of orchestration. It relates on a set of perceptual factors. It is commonly assumed to the auditory fusion of simultaneous sounds and can to concern the auditory fusion of concurrent sounds be linked to several acoustic factors (e.g., temporal into a single timbre, with the individual sounds losing synchrony, harmonicity, spectral relationships). Previ- their distinctness. Furthermore, it is thought to span ous research has left unanswered if and how musicians a perceptual continuum from complete blend to distinct control these factors during performance to achieve perception of individual timbres (Kendall & Carterette, blend. For instance, timbral adjustments could be ori- 1993; Lembke & McAdams, 2015; Reuter, 1996; Sandell, ented towards the leading performer. In order to study 1991, 1995; Tardieu & McAdams, 2012). Perceptual cues such adjustments, pairs of one bassoon and one horn that are favorable to blend range from synchronous player participated in a performance experiment, note onsets and pitch relationships emphasizing the which involved several musical and acoustical factors. harmonic series, to instrument-specific acoustical traits. Performances were evaluated through acoustic mea- Concerning pitch relationships, higher blend is sures and behavioral ratings, investigating differences achieved for unison than for non-unison intervals (Ken- across performer roles as leaders or followers, unison dall & Carterette, 1993). Whereas dissonant pitch inter- or non-unison intervals, and earlier or later segments vals exhibit greater frequency divergence between of performances. In addition, the acoustical influence harmonics that may render the identities of constituent of performance room and communication impairment instruments in a mixture more distinct, combinations in were also investigated. Role assignments affected spec- highly consonant intervals (octaves, fifths) can be tral adjustments in that musicians acting as followers assumed to be more blended. For the latter, auditory adjusted toward a ‘‘darker’’ timbre (i.e., realized by fusion can be further enhanced by parallel movement reducing the frequencies of the main formant or spec- of voices (Bregman, 1990). For all non-unison intervals, tral centroid). Notably, these adjustments occurred certain combinations of instruments can be expected to together with slight reductions in sound level, lead to higher degrees of blend than others, which may although this was more apparent for horn than bas- influence the instrumentation choices orchestrators soon players. Furthermore, coordination seemed more make. critical in unison performances and also improved With respect to acoustic traits, previous studies have over the course of a performance. These findings com- shown spectral properties to have the strongest effect on pare to similar dependencies found concerning how blend between sounds from sustained instruments. The performers coordinate their timing and suggest that global spectral shape of many wind instruments has performer roles also determine the nature of adjust- been shown to be largely invariant with respect to pitch ments necessary to achieve the common aim of and may also bear prominent features such as spectral a blended timbre. maxima (Lembke & McAdams, 2015). These maxima Received: January 27, 2015, accepted April 7, 2017. are also termed formants, in direct analogy to the pitch- independent spectral maxima found in human voice pro- Key words: ensemble coordination, music performance, duction (Fant, 1960). Previous explanations that relate timbre, blend, spectrum blend to spectral features are either based on global spec- tral characterization or focus on local, prominent spectral Music Perception, VOLUME 35, ISSUE 2, PP. 144–164, ISSN 0730-7829, ELECTRONIC ISSN 1533-8312. © 2017 BY THE REGENTS OF THE UNIVERSITY OF CALIFORNIA ALL RIGHTS RESERVED. PLEASE DIRECT ALL REQUESTS FOR PERMISSION TO PHOTOCOPY OR REPRODUCE ARTICLE CONTENT THROUGH THE UNIVERSITY OF CALIFORNIA PRESS’S REPRINTS AND PERMISSIONS WEB PAGE, HTTP://WWW.UCPRESS.EDU/JOURNALS.PHP?P¼REPRINTS. DOI: https://doi.org/10.1525/MP.2017.35.2.144 Timbre Blending During Musical Performance 145 traits. The global and more general hypothesis was the influence of performance by relying on stimuli that established from studies for instrument dyads, in which were mixed from instrument sounds that had been the spectral centroids of individual instruments were recorded in isolation, with there being a single exception evaluated. The spectral centroid represents the global, (Kendall & Carterette, 1993) in which dyad stimuli had amplitude-weighted frequency average of a spectrum. been recorded in a joint performance (Kendall & Carter- It has been shown that higher degrees of blend are ette, 1991). The interaction between performers may in obtained when the sum of the spectral centroids of fact influence blend in a way that previous research has the constituent instruments are lower (Sandell, 1995; not considered. For instance, differences between per- Tardieu & McAdams, 2012). The alternative hypothesis former roles could provide answers to the question of argues that localized spectral features influence blend, a certain instrument serving as a reference. more specifically, concerning formant relationships between instruments: when two instruments exhibit MUSICAL PERFORMANCE coincident formant locations, high blend is achieved, Psychological research on musical performance has pri- whereas increasingly divergent formant locations marily investigated temporal properties. Although past decrease blend, as the individual identities of instru- investigations have focused on note synchronization ments are thought to become more distinct (Reuter, and timing between performers (Goebl & Palmer, 1996). 2009; Keller & Appel, 2010; Rasch, 1988) as well as Lembke and McAdams (2015) followed up on the related motion cues (D’Ausilio et al., 2012; Goebl & formant hypothesis by studying frequency relationships Palmer, 2009; Keller & Appel, 2010), performer coordi- between the most prominent main formants. The inves- nation with respect to timbral properties remains tigation considered dyads of recorded and synthesized largely unexplored (Keller, 2014; Papiotis, Marchini, instrument sounds. The recorded sound remained Perez-Carrillo, & Maestre, 2014). Rasch (1988) estab- a static reference and the synthesized sound was varied lished that a certain degree of asynchrony between per- parametrically with respect to its formant frequency. formers is common and practically unavoidable, For the instruments with prominent formant structure, whereas perceptual simultaneity between musical notes namely bassoon, (French) horn, trumpet, and oboe, is still conveyed. For example, typical asynchronies blend was found to decrease markedly when the syn- between wind instruments (e.g., single and double reed) thesized main formant exceeded that of the reference, performing in non-unison are reported as falling within whereas comparably high degrees of blend were 30-40 ms. Moreover, the asynchronies relate to different achieved if the synthesized formant remained at or roles assumed by musical voices (e.g., the melody gen- below the reference. This rule proved to be robust across erally precedes bass and middle voices). different pitches, with the exception of the highest Two studies investigated the relationship between two instrument registers, and even applied to non-unison pianists being assigned performer roles as either leader pitch intervals. However, this rule relies on one instru- or follower. In one study, followers exhibited delayed ment serving as a reference, which raises the conun- note onsets relative to leaders (Keller & Appel, 2010), drum of which of two instruments in an arbitrary whereas in the other, followers displayed a higher tem- combination would function as the reference. The poral variability, thought to be linked to a strategy of answer may lie in musical practice: either the instru- error correction relative to leaders (Goebl & Palmer, ment leading the joint performance or the one with 2009). In addition, the second study showed that under a more dominant timbre could assume this function. impaired acoustical feedback, performers increasingly In musical practice, achieving blended timbres relied on visual cues to maintain synchrony. Investiga- involves two stages: its conception and its realization. tions with a sole focus on performance-related factors Blend is first conceived by composers and orchestrators, within the auditory domain would therefore need to who lay out the foundations by providing necessary per- prevent visual communication between musicians. ceptual cues, i.e., ensuring that musical parts have syn- Role dependencies between performers are indeed chronous note onsets and pitch relationships favorable to common to performance practice. They have been inves- blend, with the parts being assigned to suitable instru- tigated for larger ensembles (D’Ausilio