Real-time DSP-based audio pattern recognition system

Authors

  • Robert dela Cruz Microsystems Science, Inc., Pasig City
  • Vincent Daria Microsystems Science, Inc., Pasig City
  • Hilary de Leon Microsystems Science, Inc., Pasig City

Abstract

Advances in digital signal processing has brought about new topics of concern specially in the real-time analysis and recognition of certain audio signal patterns. Correlation and wavelet analysis techniques used in template pattern recognition requires a great deal of computation time and therefore can not be implemented in real-time. Variations in amplitude or audio volume also impose a problem if direct correlation is used. In order to avoid this problem, the audio pattern recognition algorithm is made amplitude-independent by operating in the frequency domain. Assuming that the plot between the dominant frequency and time (or the pitch plot) of a particular audio signal is unique, a signature pitch plot can then be acquired. Templates of these frequency plots could then be used to identify these audio patterns as their signature.

Downloads

Issue

Article ID

SPP-1996-IC-09

Section

Instrumentation and Computational Physics

Published

1996-12-06

How to Cite

[1]
R dela Cruz, V Daria, and H de Leon, Real-time DSP-based audio pattern recognition system, Proceedings of the Samahang Pisika ng Pilipinas 14, SPP-1996-IC-09 (1996). URL: https://proceedings.spp-online.org/article/view/SPP-1996-IC-09.