Automatic transcription of Turkish microtonal music

Abstract

Automatic music transcription, a central topic in music signal analysis, is typically limited to equal-tempered music and evaluated on a quartertone tolerance level. A system is proposed to automatically transcribe microtonal and heterophonic music as applied to the makam music of Turkey. Specific traits of this music that deviate from properties targeted by current transcription tools are discussed, and a collection of instrumental and vocal recordings is compiled, along with aligned microtonal reference pitch annotations. An existing multi-pitch detection algorithm is adapted for transcribing music with 20 cent resolution, and a method for converting a multi-pitch heterophonic output into a single melodic line is proposed. Evaluation metrics for transcribing microtonal music are applied, which use various levels of tolerance for inaccuracies with respect to frequency and time. Results show that the system is able to transcribe microtonal instrumental music at 20 cent resolution with an F-measure of 56.7%, outperforming state-of-the-art methods for the same task. Case studies on transcribed recordings are provided, to demonstrate the shortcomings and the strengths of the proposed method.

Computational Analysis of Turkish Makam Music: Review of State-of-the-Art and Challenges

“…and the name of the çeşni (Buselik, Nişabur, Hicaz, etc.) For example, in the first segmented phrase (labeled as Buselik in Neva), the resolution is on note neva (D), the chord, used to create the melodic phrase, composed of the notes neva (D), hüseyni (E) and acem (F) is the Buselik tri-chord 17 and nim hicaz (C#) note functions as a leading tone for this chord. In the computational analysis literature, to the best of our knowledge, no study performs or targets such an analysis for Turkish makam music.…”

Automatic Transcription of Polyphonic Vocal Music

“…Notice that the false positive pitch detection [41,3] is not assigned to any hypothesis state since its values of P t (p|v) are relatively small. Meanwhile, the P t (p|v) estimates from the PLCA for the other two pitches are quite good and allow the HMM to correct itself (assuming good parameter settings), judging that the voice { [43,1], [42,2], [43,3]} in the higher voice is more likely than the voice { [40,1], [42,2], [40,3]} in the lower voice, even given the noisy P t (p|v) estimates for the note [42,2].…”

“…As time-frequency representation, we use a normalised variable-Q transform (VQT) spectrogram [39] with a hop size of 20 ms and 20-cent frequency resolution. For convenience, we have chosen a pitch resolution that produces an integer number of bins per semitone (five in this case) and is also close to the range of just noticeable differences in musical intervals [40]. The input VQT spectrogram is denoted as X ω,t ∈ R Ω×T , where ω denotes log-frequency and t time.…”

Fundamental frequency estimation for monophonical Turkish music by using VMD

“…Otomatik olarak notaya dökme problemi uluslararası alanda uzun süredir üzerinde çalışılan bir konu olmasına ragmen, Türk müzigi eserleri için bu tür çalışmalar henüz çok yetersizdir. Batı müzigi için geliştirilmiş olan otomatik notaya dökme teknikleri, iki müzik arasındaki farklar nedeniyle Türk makam müzigine uygulanamamaktadır [1]. Bunun yanı sıra, Türk müzigi icrası çok fazla süsleme içermektedir.…”