Salt Lake City, Utah
June 20, 2004
June 20, 2004
June 23, 2004
9.1192.1 - 9.1192.8
Teaching Real-time DSP applications (Voice Removal) with the C6711 DSK and MATLAB
George W.P. York, Christopher M. Rondeau, Dane F. Fuller U.S. Air Force Academy, CO
This paper describes our efforts to teach real-time DSP applications at the undergraduate level. In particular this paper focuses on the voice removal DSP application, removing the lead singer from an audio recording. We find using a real-time DSP application that the students can relate to, like voice removal and other audio special effects, as a course final project highly motivates the students and makes basic DSP concepts more meaningful. While MATLAB simulations are useful for teaching the basic theory, many of these concepts are more easily taught to undergraduates if appropriate real-time demonstrations and laboratory experiences are available. The challenge of transitioning from MATLAB to real-time hardware is often the expense and a steep learning curve for the students. This paper describes a real-time DSP educational platform based around the programming ease of MATLAB and the low-cost Texas Instruments C6711 digital signal processing starter kit. Classroom uses of this platform are discussed.
While there are many interesting real-time audio DSP applications to choose from, we have found the relatively simple application of voice removal to intrigue the students and offer many engineering tradeoffs, teaching many real-time DSP issues along the way. Voice removal is attempting to remove the lead singer’s voice from an audio recording while maintaining the background instruments, useful to make a low-cost karaoke machine. We will first describe the details of a simple voice removal algorithm, then discuss our teaching paradigm, and methods used to easily transition the students from designing in MATLAB, to pseudo-real-time implementation with MATLAB running on the C6711 DSK, to real-time implementation in the C language on the DSK.
2. Voice Removal Algorithm
Voice removal works on the assumption that the lead singer’s voice is equally recorded in both the left and right channels (i.e., center stage), while the background instruments and background singers have a phase shift in either the left or right channel (i.e., stage left or stage right)1. This assumption usually does not hold for bass frequencies in which all signals tend to be equally recorded in both the left and right channels. Therefore, as shown in Figure 1, the voice removal application must first separate the bass from the rest of the signal. Then the higher frequency signals on the left and right channels are subtracted, resulting in a new signal where the
Proceedings of the 2004 American Society for Engineering Education Annual Conference & Exposition Copyright 2004, American Society for Engineering Education
York, G. (2004, June), Teaching Real Time Dsp Applications (Voice Removal) With The C6711 Dsk And Matlab Paper presented at 2004 Annual Conference, Salt Lake City, Utah. 10.18260/1-2--13979
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: © 2004 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