SYNTHESIS OF TURKISH MAKAM MUSIC SCORES USING AN ADAPTIVE TUNING APPROACH Hasan Sercan Atlı, Sertan S¸enturk¨ Xavier Serra Music Technology Group Barıs¸Bozkurt Music Technology Group Universitat Pompeu Fabra University of Crete Universitat Pompeu Fabra fhasansercan.atli, barisbozkurt0 xavier.serra sertan.senturkg @gmail.edu @upf.edu @upf.edu ABSTRACT insufficient or flawed. In parallel, the mainstream notation editors are currently Music synthesis is one of the most essential features of mu- sic notation software and applications aimed at navigating digital designed for Eurogenetic musics. While these editors pro- music score libraries. Currently, the majority of music synthesis vide a means to compose and edit music scores in Western tools are designed for Eurogenetic musics, and they are not able notation (and sometimes in other common notation formats to address the culture-specific aspects (such as tuning, intonation such as tablatures), the synthesis solutions they provide are and timbre) of many music cultures. In this paper, we focus on the tuning dimension in musical score playback for Turkish Makam typically designed for 12 tone-equal-tempered (TET) tun- Music (TMM). Based on existing computational tuning analysis ing system, and they have limited support to render inter- methodologies, we propose an automatic synthesis methodology, mediate tones and mictotonal intervals. The wide use of which allows the user to listen to a music score synthesized ac- these technologies may negatively impact the music cre- cording to the tuning extracted from an audio recording. As a ation process by introducing a standardized interpretation proof-of-concept, we also present a desktop application, which allows the users to listen to playback of TMM music scores ac- and it might even lead to loss of some variations in the ex- cording to the theoretical temperament or a user specified refer- pression and understanding of the music culture in the long ence recording. The playback of the synthesis using the tuning term (McPhail, 1981; Bozkurt, 2012). extracted from the recordings may provide a better user expe- For such cases, culture-specific information inferred rience, and it may be used to assist music education, enhance music score editors and complement research in computational from music performances may significantly improve mu- musicology. sic score synthesis by incorporating the flexibility inherent in interpretation. In this study, we focus on the tuning and 1. INTRODUCTION temperament dimensions in music score synthesis, specif- ically for the case of Turkish makam music (TMM). Turk- A music score is a symbolic representation of a piece of ish makam music is a suitable example since performances music that, apart from the note symbols, it contains other use diverse tunings and microtonal intervals, which vary information that helps put those symbols into proper con- with respect to the makam (melodic structure), geograph- text. If the score is machine-readable, i.e. the elements ical region and artists. Based on an existing computa- can be interpreted by a music notation software, the dif- tional tuning analysis methodology, we propose an adap- ferent musical elements can be edited and sonified. This tive synthesis method, which allows the user to synthesize sonification can be done using a synthesis engine and with the melody in a music score either according to a given it, the users get an approximate real-time aural feedback tuning system or according to the tuning extracted from on how the notated music would sound like if played by a audio recordings. In addition, we have developed a proof- performer. of-concept desktop application for the navigation and play- Currently, most of the music score synthesis tools ren- back of the music scores of TMM, which uses the adaptive der the audio devoid of the performance added expression. synthesis method we propose. To the best of our knowl- It can be argued that this process provides an exemplary edge, this paper presents the first work on performance- rendering reflecting theoretical information. However, the driven synthesis and playback of TMM. music scores of many music cultures do not explicitly in- For reproducibility purposes, all relevant materials such clude important information related to performance aspects as musical examples, data and software are open and pub- such as the timing, dynamics, tuning and temperament. licly available via the companion page of the paper hosted These characteristics are typically added by the performer, in the Compmusic Website. 1 by using his or her knowledge of the music, in the con- The rest of the paper is structured as follows: Section text of the performance. Some aspects of the performance, 2 gives a brief information of TMM. Section 3 presents such as the tuning and temperament, may differ due to the an overview of the relevant commercial music synthesis musical style, melodic context and aesthetic concerns. In software and the academic studies. Section 4 explains the performance-driven music styles and cultures, the “theo- retical“ rendering of a music score might be considered as 1 http://compmusic.upf.edu/node/339 Proceedings of the 7th International Workshop on Folk Music Analysis, 14-16 June 2017, Málaga (SPAIN) 66 ISBN: 978-84-697-2303-6 methodology that adapts the frequencies of the notes in a Name Flat Sharp Hc machine readable music score to be synthesized and the Koma 1 preparation of the tuning presets. Section 5 explains the Bakiye 4 music score collection, the implementation of the method- Kuc¸¨ uk¨ mucennep¨ 5 ology and the desktop software developed for discovering Buy¨ uk¨ mucennep¨ 8 the score collection. Section 6 wraps up the paper with a brief discussion and conclusion. Table 1: The accidental symbols defined in extended West- ern notation used in TMM, their theoretical intervals in Hc according to the AEU theory. 2. TURKISH MAKAM MUSIC 3. BACKGROUND Most of the melodic aspects of TMM can be explained by the term makam. Each makam has a particular scale, Many commercial music notation software tools such as which gives the “lifeless” skeleton of the makam (Signell, Sibelius 3 , Finale 4 and MuseScore 5 support engraving 1986). Makams are modal structures (Powers, et al., 2013), and editing the accidentals used in Turkish makam music. which gains its character through its melodic progression However, they provide no straightforward or out-of-the- (seyir in Turkish) (Tanrıkorur, 2011). Within the progres- box solution for microtonal synthesis. For example, Mus- sion, the melodies typically revolve around an initial tone eScore only supports synthesis of 24 tone-equal-temperament (bas¸langıc¸ or guc¸l¨ u¨ in Turkish) and a final tone (karar in system, which is not sufficient to represent the the intervals Turkish) (Ederer, 2011; Bozkurt et al., 2014). in either TMM practice or theory. Karar is typically used synonymous to tonic, and the Mus2 6 is a music notation software specifically de- performance of a makam ends almost always on this note. signed for the compositions including microtonal content. There is no definite reference frequency (e.g. A4 = 440Hz) It includes a synthesis tool that allows users to playback to tune the performance tonic. Musicians might choose to music scores in different microtonal tuning systems such as perform the music in a number of different transpositions just intonation. In addition, Mus2 allows the users to mod- (ahenk in Turkish), any of which might be favored over ify the intervals manually. Nevertheless, manually spec- others due to instrument/vocal range or aesthetic concerns ifying the intervals could be tedious. In addition, the pro- (Ederer, 2011). cess may not be straightforward for many users, which do There are several theories attempting to explain the ma- not have a sufficient musical, theoretical or mathematical kam practice (Arel, 1968; Karadeniz, 1984; Ozkan,¨ 2006; background. Yarman, 2008). Among these, Arel-Ezgi-Uzdilek (AEU) There exists several studies in literature for automatic theory (Arel, 1968) is the mainstream theory. AEU the- tuning analysis of TMM (Bozkurt, 2008; Gedik & Bozkurt, ory is based on Pythagorean tuning (Tura, 1988). It also 2010) and Indian art musics (Serra` et al., 2011; Koduri presents an approximation for intervals by the use of Holde- et al., 2014). These studies are mainly based on pitch rian comma (Hc) 2 (Ederer, 2011), which simplifies the histogram analysis. Bozkurt et al. (2009) analyzed the theory via use of discrete intervals instead of frequency ra- recordings of masters in 9 commonly performed makams tios. “Comma” (koma in Turkish) is part of daily lexicon by computing a pitch histogram from each recording and of musicians and often used in education to specify inter- then detecting the peaks of histograms. Considering each vals in makam scales. Some basic intervals used in AEU peak as one of the performed scale degrees, they compared theory are listed in Table 1 (with sizes specified in commas the identified scale degrees with the theoretical ones de- on the last column) fined in several theoretical frameworks. The comparison th showed that the current music theories are not able to ex- Since early 20 century, a score representation extend- plain the intervallic relations observed in the performance ing the traditional Western music notation has been used as practice well. a complement to the oral practice (Popescu-Judetz, 1996). Later, Bozkurt (2012) proposed an automatic tuner for The extended Western notation typically follows the rules TMM. In the tuner, the user can specify the makam and in- of AEU theory. Table 2 lists the accidental symbols spe- put an audio recording in the same makam. Then, the tun- cific to TMM used in this notation. ing is extracted from the audio recording using the pitch The music scores tend to notate simple melodic lines histogram analysis method described above. The tuning and the musicians follow the scores of the compositions as information is then provided the user interactively, while a reference.