Aligning Audio and EGG: Polarity and Delay

Note: All terms appearing in bold letters areĀ found in the VoceVista displays.

An essential step toward a coordinated analysis of the mic and EGG signals is to align the two signals in time so that closing and opening are precisely synchronous. This involves the correction of a delay: the impulse produced by the closing of the glottis, which is registered directly in the EGG, arrives at the microphone only after a journey through the air. This delay is on the order of 1.5 ms for a microphone 30 cm (12 inches) from the subject’s mouth, or 0.75 ms for a head-mounted microphone.

A second point in analyzing the two signals concerns polarity. The EGG signal should be displayed so that the sharp increase in vocal-fold contact that represents glottal closing is an upward movement, and the more gradual slope of glottal opening is downward. Determining the polarity of the mic (Audio) signal is less straightforward. In the most characteristic case, the abrupt closing of the glottis is reflected in a sudden rise in this signal, following a segment with a somewhat flatter, usually falling, curve of the open phase, which is generally the weakest acoustic segment of the glottal cycle.

The figure shows an example from the lower part of the male range. The Delay (reading 0.00 ms) has not yet been adjusted, and the Polarity of both signals is incorrect. The evidence of glottal closing in both signals is apparent, however, and is indicated with arrows in the figure.


To correct the display, go to Analysis and reverse both Audio Polarity and EGG Polarity. Using the left/right arrows, now move the audio signal so that the outside vertical cursors pass directly through the beginning of the nearly vertical segments marking glottal closing in both Audio and EGG. The Delay should be something close to 0.90 ms in this particular case. A second figure shows the same signals, now corrected for polarity and delay:


Once the Delay has been correctly set, that setting can be considered valid for subsequent measurements so long as the microphone distance is kept constant. With the delay set one can also scroll through the waveform display as slowly as 1 ms per click by going to Analysis/ Arrow Key Assignment and clicking on Toggle L/R. This will couple the audio and EGG signals so that they move together at a fixed delay. Subsequent adjustments of delay will require a reversal of the toggle switch.

Locating the moment of closure in the audio waveform is relatively easy when the pitch is low and the vowel is open (high first formant). Thus it is helpful to include such tones within a given phonation in order to establish correct delay.

Adjusting the delay for the higher F0s, particularly for female voice, can be more challenging. The Exercises contain a similar example with a soprano voice.

Audio Segment Length

VoceVista will display between 5 and 50 ms of Audio and EGG signals, adjustable under Analysis Options. If the displayed segment is not long enough to include two full glottal cycles, the EGG waveform may fail to appear, and the closed quotient history (EGG CQ) will not reflect the true closed quotient. This can be corrected simply by increasing the Audio Segment Length.

An optimal setting of the Audio Segment Length would display about 2 1/2 cycles on the lowest pitches of the phonation. The following figures show a bass singing a phrase that descends from B3 to B2. The first figure has a setting of 12 ms for the Audio Segment Length — too short to display the low pitches properly. Note especially what happens to the CQ history (EGG CQ) as the pitch descends (arrow).


In the second figure the Audio Segment Length has been increased to 20 ms, and the CQ history is restored.


For female singing in higher range, a setting of 20 ms will be too high for ideal display. For close examination of notes above the staff, for example, the minimum of 5 ms is appropriate. The following example shows two moments (pp and ff, respectively) in a female messa di voce at the pitch A5-flat.


The maximum of 50 ms is useful for displaying the details of a rapid change. It is used for the following figure, which shows a female singer ending a sustained F4 (358 Hz in display) with a glottal stop. The CQ history shows an increase in CQ as the stop is approached (left arrow), and the 50 ms display of the EGG signal shows the final irregular vibrations of the vocal folds after periodic vibration has ceased (right arrow). The closed quotient of the second cycle displayed in the right hand panels, one of the last cycles before vibration ceases, is measured as 61%.


For better detail in the time dimension, the right-hand panels can be doubled in length by clicking on the F8 display mode.


Configuring VoceVista