Melody Extraction
Melody extraction is the task of automatically isolating the main melodic contour from polyphonic music recordings. It originated from music transcription research in the 2000s and addresses the core challenge of human pitch perception: identifying the perceptually dominant pitch when many instruments play simultaneously. Modern approaches use deep learning and are essential for music analysis, cover song detection, and music-to-lyrics alignment.
Source record
Citations copied verbatim from the method’s source record. No claim-level verification is inferred from them.
- Salamon, J., & Gómez, E. (2014). Melody extraction from polyphonic music signals using pitch contour characteristics. IEEE Transactions on Audio, Speech, and Language Processing, 20(6), 1759-1770. · URL
- Klapuri, A. (2008). Automatic music transcription as we know it today. Journal of New Music Research, 33(3), 323-337. · DOI 10.1007/978-0-387-30441-0_20
- Bittner, R. M., McVicar, M., Salamon, J., & Ellis, D. P. (2017). An analysis of lead and accompaniment separation in polyphonic music. In Proceedings of the International Society for Music Information Retrieval Conference. · URL
Curated claims
Claims persisted in the evidence ledger, each with its own assessment.
This view does not invent a claim assessment when the ledger has none.
Related methods
Generated from the method graph and shown as machine-suggested relations — no evidence claim is inferred.