Asee peer logo

A Speech And Music Detector Project For A Dsp Class

Download Paper |

Conference

2001 Annual Conference

Location

Albuquerque, New Mexico

Publication Date

June 24, 2001

Start Date

June 24, 2001

End Date

June 27, 2001

ISSN

2153-5965

Page Count

7

Page Numbers

6.101.1 - 6.101.7

Permanent URL

https://peer.asee.org/9793

Download Count

67

Request a correction

Paper Authors

author page

Christopher Vondrachek

author page

Joseph Hoffbeck

Download Paper |

Abstract
NOTE: The first page of text has been automatically extracted and included below in lieu of an abstract

Session 2793

A Speech and Music Detector Project for a DSP Class

Christopher J. Vondrachek, Joseph P. Hoffbeck

University of Portland

Abstract

A project is described in this paper which is designed to monitor a radio station and detect commercials and talking, which would allow the radio to be muted so the listener would not be distracted by obnoxious radio ads and D.J.’s. The project is designed to be an interesting application of a very simple pattern recognition system and requires little more than a low pass filter, high-pass filter, and a threshold scheme. The approach was to attempt to classify the signal as music or speech, and if the signal was found to be speech, the device could mute the radio (advertisements with music would not be eliminated). The device was based on the TMS320C31 DSK, which is an inexpensive digital signal processor (DSP) demonstration board available from Texas Instruments that contains an A/D converter, D/A converter, and a computer interface. The DSP on the demonstration board was programmed to take samples from the A/D converter, pass them through a low-pass and high-pass filter, and to compute the ratio of the average energy at the output of the LPF filter to the average energy of the output of the high pass filter. Since, most of the time, speech has less energy at high frequencies than music does, a high ratio usually indicates speech. The DSP compares the ratio to a threshold value, which was determined from recorded speech and music, and, if the ratio is higher than the threshold, generates a signal that could be used to mute the speech. The technique was found to work well with the test signals used to find the threshold, but did not generalize very well to new music and speech signals.

I. Introduction

The challenge in many courses is not in deciding how best to teach a topic, but in deciding how best to motivate students so that they want to learn the material. The project described in this paper is designed to show an interesting and useful application of DSP that requires little theory, and can be used to capture the interest of the students and motivate them to learn DSP theory. It shows how a low pass filter, a high pass filter, and a thresholding scheme can be used to determine if an audio signal is music or speech so that a radio could automatically mute commercials and other talking. Using the inexpensive TMS320C31 DSK, this simple speech/music recognition algorithm can be run and demonstrated in real time, with real music and speech signals.

The project can be used in courses in several different ways. For example, the students could analyze speech and music samples to determine the filter coefficients and the value of the

Proceeding of the 2001 American Society for Engineering Education Annual Conference & Exposition Copyright 2001, American Society for Engineering Education

Vondrachek, C., & Hoffbeck, J. (2001, June), A Speech And Music Detector Project For A Dsp Class Paper presented at 2001 Annual Conference, Albuquerque, New Mexico. https://peer.asee.org/9793

ASEE holds the copyright on this document. It may be read by the public free of charge. Authors may archive their work on personal websites or in institutional repositories with the following citation: © 2001 American Society for Engineering Education. Other scholars may excerpt or quote from these materials with the same citation. When excerpting or quoting from Conference Proceedings, authors should, in addition to noting the ASEE copyright, list all the original authors and their institutions and name the host city of the conference. - Last updated April 1, 2015