Our experiments involve subjects producing a vocalization, usually a vowel (ah) screen-shot-2016-12-01-at-9-48-54-amfor several seconds. During the vocalization, the subject’s voice pitch feedback is suddenly shifted, which results in the subject automatically changing his/her voice fundamental frequency (F0) output. Since the subject’s response is generally quite small, several trials are averaged together to produce the averaged response, as shown below and in the following information.

Generating an F0 Contour 

Because the subject’s voice is represented graphically as an alternating signal, we must convert this signal into a waveform in which the voice F0 is represented as an analog trace, or contour, as shown below.  To do this, we measure the time between each glottal cycle and take the reciprocal of this number, which is the frequency.  We then plot these values as a function of time, which results in the F0 contour.

Segmenting Data

Following the process of converting the audio waveform into F0 contours, the contours are segmented into individual trials with each pitch-shift stimulus.



Averaging Voice F0 Contours

Following the process of converting the audio waveform into F0 contours and segmenting them into individual trials with each pitch-shift stimulus, the average of the group of individual contours is calculated (middle graph below).


Average Examples

Examples of average responses to upward (top) and downward (bottom) pitch-shift stimulus.  In both cases, the response is in the opposite direction to the stimulus and are called “Compensating Responses”. Also shown in this figure are control trials in which no pitch-shift stimulus was presented.


Example of a compensating response (above) and a following  response  (below). Following responses are fewer in number; these change in the same direction as the stimulus.


EEG Signals from Multiple Electrodes

Another technique we use is to record brain electrical signals during the pitch-shift paradigm in an effort to understand the neural mechanisms underlying these processes. An electrode cap consisting of 32 or 64 EEG electrodes is placed on the subject’s head.  During the testing as the subject is vocalizing and hearing his/her voice pitch shifted, electrodes pick up tiny electrical signals from the brain.  Recordings from the electrodes are synchronized with the pitch-shift stimulus.  After the recording session is completed, the electrical potentials from each electrode are averaged over the course of all the pitch-shifts, resulting in an Event Related Potential (ERP).


Examples of Event Related Potentials (ERPs):


Brain Event-Related Potentials from EEG Electrodes

Below are averages of ERPs recorded from musicians with perfect pitch, relative pitch and non-musicians during a pitch-shift study. The larger amplitude of the P200 ERP in perfect pitch musicians may reflect neural mechanisms that endow them with unique musical abilities.