Abstract
Interactive musical systems require real-time, low-latency, accurate, and reliable event detection and classification algorithms. In this paper, we introduce a model-based algorithm for detection of percussive events and test the algorithm on the detection and classification of different percussive sounds. We focus on tuning the algorithm for a good compromise between temporal precision, classification accuracy and low latency. The model is trained offline on different percussive sounds using the expectation maximization approach for learning spectral templates for each sound and is able to run online to detect and classify sounds from audio stream input by a Hidden Markov Model. Our results indicate that the approach is promising and applicable in design and development of interactive musical systems.
Highlights
Percussion instruments traditionally provide the rhythmic backbone in music
We have witnessed an increase in the number of interactive applications built around sound detection and classification algorithms, which has been enabled by the increase in computational power of computers and appliances
We explore the domain of realtime automatic detection and labeling of percussive sounds for interactive systems
Summary
Percussion instruments traditionally provide the rhythmic backbone in music. In the past few years, their automatic detection and classification has been studied in the context of music information retrieval for numerous purposes, including metrical analysis, database labeling and searches, automatic transcription, and interactive musical systems. We have witnessed an increase in the number of interactive applications built around sound detection and classification algorithms, which has been enabled by the increase in computational power of computers and appliances. While the computational power enables the use of more and more complex algorithms, the applications still favor algorithmic simplicity and demand low-latency solutions in order to make the interaction fluent. We explore the domain of realtime automatic detection and labeling of percussive sounds for interactive systems. These systems require an efficient and reliable low-level method for sound analysis. Our solution to the problem is a probabilistic model-based algorithm, which efficiently detects and labels percussive events. The realization of a real-time object capable of labeling the events in an audio stream is discussed, with special attention to its applicability in interactive systems
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call