FUGA-ID: A Cross-Modal Approach to Symbolic Folk Melody Identification

This archive contains the complete codebase, symbolic data, and evaluation materials used in the development of FUGA-ID: A Cross-Modal Approach to Symbolic Folk Melody Identification. It includes the FUGA-ID retrieval system, the Folkoteca Galega Research Corpus (FGRC) symbolic dataset, and two sets of audio recordings created specifically for this project. The symbolic component includes original scores provided by the Folkoteca Galega, which is the collector and custodian of the traditional notated materials. The audio recordings and performance-based data were produced expressly for the development and evaluation of the FUGA-ID system. Contents FUGA-ID source code for cross-modal melody identification. Folkoteca Galega symbolic dataset: 2,116 traditional Galician melodies in MusicXML (2,083 also in MIDI), covering 25 genres. Evaluation recordings including: A Multi-Instrument Set (156 WAV files across 8 instruments). A Piano Dual-Capture Set (50 performances in WAV and MIDI). Multiple recordings of the same melodies, enabling cross-modal and cross-performance retrieval experiments. Feature extraction and indexing tools. Reproducible benchmarking pipeline, AMT evaluation scripts, similarity analysis, and performance metrics. Results database and analysis utilities for generating reports and exploring retrieval outcomes. Purpose These materials are provided to ensure transparent evaluation, reproducibility, and easy access to symbolic and audio data for research on folk melody identification, AMT benchmarking, and melodic similarity analysis.