Aggregate features and ADABOOST for music classification
Machine Learning
Musical instrument recognition using cepstral coefficients and temporal features
ICASSP '00 Proceedings of the Acoustics, Speech, and Signal Processing, 2000. on IEEE International Conference - Volume 02
Musical instrument timbres classification with spectral features
EURASIP Journal on Applied Signal Processing
EURASIP Journal on Applied Signal Processing
A New Classifier to Deal with Incomplete Data
SNPD '08 Proceedings of the 2008 Ninth ACIS International Conference on Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing
A Multipitch Analyzer Based on Harmonic Temporal Structured Clustering
IEEE Transactions on Audio, Speech, and Language Processing
Hi-index | 0.00 |
Musical Instrument Identification research is an important problem in Music Information Retrieval (MIR) in which most of the research going on now is using signal processing method. In this paper, at first a model that uses harmonic structured Gaussian mixture for modeling instrument is described and EM algorithm is used to estimate parameters in the model. Therefore features such as Harmonic Temporal Timbre Energy Ratio (HTTER) and Harmonic Temporal Timbre Envelope Similarity (HTTES) are generated from the model. To utilize the features efficiently, a new boosting algorithm based on Probabilistic Decisions is proposed for musical instrument identification. In contrast to the conventional boosting algorithm, which uses a deterministic decision method during the iterations and which does not consider the noise in the data set sufficiently, the new boosting algorithm is proposed to use probabilistic decisions for every hypothesis at the iterations of the boosting scheme, selecting the data events from a dataset, and then combines them. It improves the musical instrument classifier without using boosting approach and the conventional boosting algorithm significantly, which was proved by the experimental.