Skip to main content

Computational Analysis Of Traditional Georgian Vocal Music (GVM) [DFG MU 2686/13-1, SCHE 280/20-1]

A central aspect of the GVM project was the investigation of the tonal organization of traditional Georgian vocal music. This topic has been the subject of intense, sometimes very controversial scholarly discussions for many decades. In contrast to previously existing studies, the GVM project approached it from a purely descriptive, data-driven perspective. In order to ensure maximum transparency of this work, all analyses were documented comprehensively and made publicly available together with the raw data. In total, two large audio data sets were analyzed, the so-called Erkomaishvili dataset (Rosenzweig et al, 2020) and the GVM dataset (Scherbaum et al., 2019).  The first one is a collection of unique recordings of liturgical chants sung by the Georgian master singer Artem Erkomaishvili in 1966. Due to the time and setup of its recording, this corpus is of outstanding importance for the understanding of the tuning principles of traditional Georgian vocal music. The GVM dataset is a collection of high-quality field recordings, obtained during the preparatory phase of the GVM project. Both datasets were systematically and comprehensively analyzed using modern state-of-the-art pitch analysis algorithms.

The analysis of the Erkomaishvili dataset is documented in the journal papers Scherbaum et al. (2017), Müller et al. (2017),  Rosenzweig et al. (2020), Scherbaum et al. (2020) and the video presentation Scherbaum et al. (2021).  

The analysis of the GVM dataset is described in, Scherbaum et al. (2018a), Scherbaum et al. (2018b),  Scherbaum et al. (2019), and  Scherbaum et al. (2022).

The second major contribution of the GVM project was a comprehensive interdisciplinary study of three-part funeral songs (aka Zär) from Svanetia in northwestern Georgia. This study was based entirely on a subset of the  GVM dataset and is described in the journal papers  Mzhavanadze and Scherbaum (2020), Scherbaum and Mzhavanadze (2020), Mzhavanadze and Scherbaum (2021), Scherbaum and Mzhavanadze (2021), and Rosenzweig et al. (2022), as well as in a video presentation at the Annual Meeting of the Society of Ethnomusicology, Ottawa, 2020 (Mzhavanadze and Scherbaum, 2020)

The third focus of the GVM project was the development of tools for computer-aided signal processing and music information retrieval (MIR) specifically designed for the analysis of multitrack vocal recordings. With reproducible research in mind, a set of well-documented and user-friendly toolboxes was developed that integrate many of these computational tools under an open-source license and provide reference implementations.  A selected subset of these are documented in Rosenzweig et al. (2019)Rosenzweig et al. (2021), and Scherbaum et al. (2023).

It is  also worth mentioning that a yet unpublished study on singer interaction was presented at the  Musical Togetherness Symposium (MTS-22), 13-15 July 2022, University of Music and Performing Arts Vienna, Austria (Scherbaum and Müller, 202.  

Finally, a brief overview  of some of the  highlights of the whole GVM project  can be obtained from the video of the talk presented by Frank Scherbaum at the 11. International Symposium on Traditional Polyphony, Tbilisi, 26- 30 September 2022. More details can also be found at the  Erlangen project page