ECHODamp was created and developed by Dr. Brian K. Shepard, Assistant Professor of Pedagogical Technology at the University of Southern California Flora L. Thornton School of Music. In October 1999, Dr. Shepard conducted the very first demonstration of a high-bandwidth, private music lesson using uncompressed, bidirectional video and audio over Internet2. As a pioneer, and one of the earliest proponents of high-bandwidth videoteleconferencing for musical purposes, he has devoted more than 10 years of research to sound-quality issues in the musical VTC. Brian has presented his research and conducted workshops around the world on audio quality and echo control in the musical VTC using a sophisticated array of audio and network components. Unfortunately, the expense and complexity of that equipment often discourages people from using his proven techniques. Thus, in an attempt to simplify the process and make high-quality audio and echo control more universally available, Dr. Shepard began developing and writing a software application that combines all of his techniques into one easy-to-use program. ECHODamp is the result of that research and work.
By definition, a VTC requires both microphones and loudspeakers at each location. Unfortunately, having live microphones in the same room with loudspeakers often causes an audio loopback when sound coming from the remote site through the speakers gets picked up by the microphones and then transmitted back to the original source where it is heard as an echo. Even with the fastest VTC codecs available, the delay on the echo can be in the range of a half second or more. For participants in the VTC, that echo can be an extreme distraction and annoyance, making effective communication and learning nearly impossible.
Over the years, a number of echo cancellers have been developed for VTCs. These devices typically work by filtering out the upper-mid and high frequencies, where much of the echo associated with speech occurs, and by using a concept known as Acoustic Echo Cancellation (AEC). AEC compares the incoming and outgoing audio streams and looks for acoustic matches. When it detects a match, it attempts to remove that sound from the audio stream with a phase-cancellation process that temporarily removes frequencies associated with the echo. For speech purposes, these two techniques can work fairly well. Speech is often made more intelligible in electronic devices when the upper, sibilant frequencies are reduced, and since speech is primarily short bursts of sound rather than long continuous tones, it is relatively simple to look for, and find, matches in the incoming and outgoing signals.
The musical VTC, however, is a completely different type of sonic environment. When musicians evaluate the sound of a musical performance, they are evaluating much more than just whether the right notes are being played. They are listening for the tone and timbre of the instrument or voice and the way it projects in the room in which it is sounding. This type of sonic information is primarily contained in the very same frequency range the typical echo canceller eliminates. Thus, it is difficult, if not impossible, to accurately evaluate a musician's sound in a VTC using traditional echo cancellation. Secondly, since music is frequently made up of long continuous tones, echo cancellers using AEC often get "confused" by the fact that the incoming and outgoing signals are quite similar and they begin to wildly fluctuate the level and the frequency content of the audio, causing unacceptable dips and cuts in the sound for the participants.
Rather than using phase cancellation and removing frequency content, ECHODamp listens for the directionality of a sound's source and uses that information to help prevent echo from entering the audio chain in the first place. If unwanted echo does enter the signal, it is gracefully damped in a manner that is both unobtrusive and musical. Thus, ECHODamp allows participants to experience a VTC with the full audio frequency spectrum and without echo.
ECHODamp was made possible by support from the USC Thornton School of Music and guidance from the USC Stevens Institute for Innovation. Invaluable technical and testing assistance came from colleagues at Internet2, The New World Symphony, Indiana University Purdue University Indianapolis Department of Music and Arts Technology, and Northern Illinois University School of Music.