Towards Computer-Assisted Flamenco Transcription: An Experimental Comparison of Automatic Transcription Algorithms as Applied to A Cappella Singing

Abstract

This article deals with automatic transcription of flamenco music recordings-more specifically, a cappella singing. We first study the specifics of flamenco singing and propose a transcription system based on fundamental frequency and energy estimation, which incorporates an iterative strategy for note segmentation and labeling. The proposed approach is evaluated on a music collection of 72 performances, including a variety of singers and recording conditions, and the presence or absence of percussion, background voices, and noise. We obtain satisfying results for the different approaches tested, and our system outperforms a state-of-the-art approach designed for other singing styles. In this study, we discuss the difficulties found in transcribing flamenco singing and in evaluating the obtained transcriptions, we analyze the influence of the different steps of the algorithm, and we state the main limitations of our approach and discuss challenges for future studies.

Robust Singing Transcription System Using Local Homogeneity in the Harmonic Structure

“…All the results are excerpts from the original papers [8], [23] that use the same dataset and criteria. The results attained by Rynnänen's note event model approach [6], Gómez & Bonada's method [7], a commercial system named Melotranscript [24] were cited from Molina's evaluation framework [23]. The SiPTH system has only one overall performance about COnPOff, since the authors do not mention the result on COnP and COn in their paper [8].…”

“…Rynnänen combined two probabilistic models to detect natural notes in a musicological sense [6]. More recently, Gómez and Bonada proposed an iterative note-consolidation technique using low-level features related to the pitch, duration, voicing and stability [7]. Molina presented a note segmentation method based on pitch-time hysteresis, making use of the dynamic average of the pitch curve [8].…”

Automatic Transcription of Flamenco Singing From Polyphonic Music Recordings

“…Nevertheless, the authors report mistakenly transcribed guitar notes as a main source of error and the inaccurate vocal detection a the major limitation of the system performance. The reported note transcription accuracies reported in [23] with a note f-measure of slightly below 0.4 furthermore indicate the difficulty of transcribing flamenco singing. Significantly higher performance is achieved when evaluating the same transcription algorithm on a dataset containing pop-and jazz singing excerpts.…”

“…In the context of flamenco singing, a first approach for monophonic transcription was proposed in [2]: Based on a contour simplification algorithm [27], an estimated monophonic pitch track is converted into a set of constant segments within which the absolute error between the pitch track and the fitted constant does not exceed a pre-determined threshold. A system for computer-assisted transcription of single-voiced a cappella singing recordings has been proposed in [23]: Given the absence of accompaniment, a monophonic pitch estimator based on spectral auto-correlation (SAC) represents the front-end of the system. The note segmentation stage is based on a likelihood maximisation method [24]: A dynamic programming (DP) algorithm is used to find the best among all possible notes segmentations along the entire track.…”

Analyzing Free-Hand Sound-Tracings of Melodic Phrases

“…Melodic contour is also used to describe the instrumentation of music from audio signals, for example in [36,37]. It is also interesting to note that melodic contour is used as the first step to identify musical structure in styles such as in Indian Classical music [38], and Flamenco [39].…”