VoicescapesPolyphony Lab

The GVM project

In short: In the GVM project (Computational Analysis of Traditional Georgian Vocal Music, DFG MU 2686/13-1, SCHE 280/20-1) we used signal-processing methods to find out how Georgian singers actually tune. We did not measure starting from a theory but let the recordings speak for themselves.

What it was about

The tonal organisation of Georgian vocal music has been debated intensely, sometimes controversially, for decades. The project took a purely descriptive, data-driven approach. All analyses are documented in detail and publicly available together with the raw data so that they can be checked.

Two large audio datasets were analysed:

  • the Erkomaishvili dataset, unique recordings of liturgical chants by the master singer Artem Erkomaishvili from 1966, of outstanding importance for understanding tuning principles,
  • the GVM dataset, high-quality field recordings from the preparatory phase of the project.

Three focal points

  1. Tonal organisation: both datasets were systematically analysed with modern pitch-analysis algorithms.
  2. Funeral songs (Zär) from Svaneti: an interdisciplinary study of the three-part funeral songs of north-western Georgia, based on a subset of the GVM dataset.
  3. Tools: well-documented, open-source toolboxes for analysing multitrack vocal recordings.

An as yet unpublished study on singer interaction was presented in 2022 at the Musical Togetherness Symposium (MTS-22) in Vienna. For an overview of the main results see the talk at the 11th International Symposium on Traditional Polyphony in Tbilisi (2022).

Further reading and data

The field recordings were made with larynx microphones.

The project was funded by the German Research Foundation (DFG).