This research focuses on utilizing the traditional frame drum as an input sensor for the creation of music sheets without the need for manual transcription, by training and evaluating models on a dataset containing samples from various frame drums and playing techniques. The existing dataset is extended to include recordings from different frame drums with varying attributes and playing techniques for improved generalization. CNN-based models are employed for audio sample recognition between the main hitting techniques and playing positions. Additional modules for pitch recognition and onset detection are proposed in the system architecture. The models are trained and evaluated on individually recorded samples.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=22592
(378KB)
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member Join the AES. If you need to check your member status, login to the Member Portal.
