Using Computers to Aid the Development of Human Vocal Production and Perception

Visualization of Intervals

 

We have developed a program that displays a visual representation of the intervallic relationship between an input sound and a user-chosen reference pitch.

How does it work

  1. The control window on the right side of the interface allows you to choose a starting pitch and an interval. 
  2. You then produce a sound and the program displays a Lissajous-style figure that represents the ratio of the pitch you are producing and a fundamental pitch calculated from the pitch and interval combination you have chosen.
  3. The display appears to ‘rotate’ counterclockwise when your input is below the pitch required to make your chosen interval (ratio) and clockwise when your input is above the required pitch.

 

Principal investigators: Pete Nordquist, Dr. R. Dean Ayers

Project contributors: Daisuke Kato, Lewis Nakao, Justin Corn

 

Click here to run the Interval Visualizer.

  1. Select the starting pitch you would like.
  2. Select the interval you would like to sound ascending or descending from this pitch. 
  3. Adjust the pitch you are singing until the display no longer rotates.
  4. By default, the input is filtered to produce a sinusoidal waveform.  You may disable this filter by using the control found on the menu in the options dialog box.

 

View slides describing the project.

View slides from my Acoustical Society of America presentation – June 5, 2007.

View slides from my science seminar talk – May 13, 2005.

 

Please contact Pete Nordquist with questions or comments about the interval visualizer. Thank you.

Pitch Detection

We have developed a program that takes an audio signal as input and produces a visual representation of the spectrum of frequencies present in this signal.  This frequency spectrum is also processed by an algorithm for detecting the fundamental frequency in the spectrum, commonly referred to by musicians as the pitch of the spectrum.

 

Our program also has the ability to compare the current fundamental frequency with a set of predetermined frequencies and output the frequency in the set that is closest to the input fundamental.  In addition, we output the difference between this closest frequency and the actual fundamental.

 

The technique of displaying a visual representation of the input fundamental frequency combined with visually displaying how far this frequency is from a particular reference frequency is very useful in training musicians, because it allows them to receive feedback with their vision as well as their hearing.  The technique of supporting an arbitrary set of reference frequencies allows the user to practice tuning to scales other than the common well-tempered scale.

 

A side effect of displaying the frequency spectrum is that this spectrum contains timbral information.  As a singer, I am interested in using this information to analyze vowel production.  Another possible use for this capability might be to train people who have lost or never had the ability to hear their own speech.  Having a visual representation of a correctly pronounced vowel sound to compare with visual feedback from their own pronunciation would allow them to improve the accuracy of their speech.

 

The computer program described above is written in java, has been tested on the Linux, MS Windows, and Apple Macintosh operating systems, and is available free of charge.  It requires no special hardware other than a microphone and sound card, which are standard equipment on almost all modern computers.

 

Principal investigator: Pete Nordquist

Project contributors: David Young, Ken Lindsay

 

Download PitchTest.jar, spectrum capture and visualization application written by David Young.

 

The Shakuhachi Tuner has similar functionality and is a free download, though source code is not available.