International Audio Laboratories Erlangen

Detecting Stable Regions in Frequency Trajectories for Tonal Analysis of
Traditional Georgian Vocal Music

This is the accompanying website for the following paper:

1. Sebastian Rosenzweig, Frank Scherbaum, and Meinard Müller
   Detecting Stable Regions in Frequency Trajectories for Tonal Analysis of Traditional Georgian Vocal Music
   In Proceedings of the International Society for Music Information Retrieval Conference (ISMIR): 352–359, 2019. Bibtex PDF Demo

   @inproceedings{RosenzweigSM19_StableF0_ISMIR,
   author    = {Sebastian Rosenzweig and Frank Scherbaum and Meinard M{\"u}ller},
   title     = {Detecting Stable Regions in Frequency Trajectories for Tonal Analysis of Traditional Georgian Vocal Music},
   booktitle = {Proceedings of the International Society for Music Information Retrieval Conference ({ISMIR})},
   pages     = {352--359},
   address   = {Delft, The Netherlands},
   year      = {2019},
   url-demo  = {https://www.audiolabs-erlangen.de/resources/MIR/2019-ISMIR-StableF0},
   url-pdf = {2019-ISMIR-StableF0/literature/2019_RosenzweigSM_StableRegions_ISMIR.pdf}
   }

Acknowledgements

This work was supported by the German Research Foundation (DFG MU 2686/13-1, SCHE 280/20-1). The International Audio Laboratories Erlangen are a joint institution of the Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) and Fraunhofer Institut für Integrierte Schaltungen IIS.

Abstract

While Georgia has a long history of orally transmitted polyphonic singing, there is still an ongoing controversial discussion among ethnomusicologists on the tuning system underlying this type of music. First attempts have been made to analyze tonal properties (e. g., harmonic and melodic intervals) based on fundamental frequency (F0) trajectories. One major challenge in F0-based tonal analysis is introduced by unstable regions in the trajectories due to pitch slides and other frequency fluctuations. In this paper, we describe two approaches for detecting stable regions in frequency trajectories: the first algorithm uses morphological operations inspired by image processing, and the second one is based on suitably defined binary time–frequency masks. To avoid undesired distortions in subsequent analysis steps, both approaches keep the original F0-values unmodified, while only removing F0-values in unstable trajectory regions. We evaluate both approaches against manually annotated stable regions and discuss their potential in the context of interval analysis for traditional three-part Georgian singing.

Downloads

This study is based on the Erkomaishvili Dataset, which can be downloaded here. Furthermore, the recordings from the dataset can be explored in a YouTube playlist. The manual annotations of stable regions from Section 4.1 in the paper can be downloaded here (ZIP archive).

Recordings                               Search: