Document is current

Article History

Received: 23 October 2024

Accepted: 9 February 2025

First Online: 28 March 2025

Declarations

:

: This research presents a methodology that enhances multi-pitch estimation for instrument-wise transcription. It is based on a publicly available codebase and utilizes publicly available music datasets for training and evaluation: MusicNet and mshoxxDB. The latter is entirely authored by MT, and explicit permission was obtained for the inclusion of two remixes of songs by other artists. While artist names are included in the dataset, no personal identifying information beyond publicly available details is disclosed (data privacy). mshoxxDB is made available to the research community to support further studies in the proposed areas. Our implementation is open-source to promote transparency, reproducibility, and collaborative advancement in the field.We acknowledge potential ethical considerations associated with transcription technology, such as for music teaching, unauthorized music reproduction, and cultural bias, and thus encourage responsible use of our methods in accordance with legal and ethical standards: This technology should be seen as a way to support educators and creators — for example, to provide tools for transcription verification and analysis of harmonic/melodic structures via visual representations. Appropriate licenses have been applied to use and create derivative works to ensure that future work is conducted under the same conditions.

: The authors declare that they have no competing interests.