Chih-Wei Wu

Drum transcription using partially fixed non-negative matrix factorization

In this paper, a drum transcription algorithm using partially fixed non-negative matrix factorization is presented. The proposed method allows users to identify percussive events in complex mixtures with a minimal training set. The algorithm decomposes the music signal into two parts: percussive part with pre-defined drum templates and harmonic part with undefined entries. The harmonic part is able to adapt to the music content, allowing the algorithm to work in polyphonic mixtures. Drum event times can be simply picked from the percussive activation matrix with onset detection. The system is efficient and robust even with a minimal training set. The recognition rates for the ENST dataset vary from 56.7 to 78.9% for three percussive instruments extracted from polyphonic music.

An Unsupervised Approach to Anomaly Detection in Music Datasets

This paper presents an unsupervised method for systematically identifying anomalies in music datasets. The model integrates categorical regression and robust estimation techniques to infer anomalous scores in music clips. When applied to a music genre recognition dataset, the new method is able to detect corrupted, distorted, or mislabeled audio samples based on commonly used features in music information retrieval. The evaluation results show that the algorithm outperforms other anomaly detection methods and is capable of finding problematic samples identified by human experts. The proposed method introduces a preliminary framework for anomaly detection in music data that can serve as a useful tool to improve data integrity in the future.