Tracking movement

Tracking articulator movements from X-ray motion films

  • Vocal tract profiles are traced frame by frame from the X-ray motion film
  • Tracings for selected sequences are superimposed to track the movement of individual articulators, for example:
    • The upper lip with respect to the maxilla
    • The mandible with respect to the maxilla
    • The lower lip with respect to the mandible
    • The tongue body with respect to the mandible
    • The tongue blade with respect to the tongue body
  • The composite pictures are then separated back to a series of individual pictures to be played in an animation sequence.

Tracing procedures

 Tracing X-ray profiles
  • Each frame is scanned at high resolution from the film, using a 35mm film scanner
  • A scan is opened in a drawing program
  • The X-rayed centimetre scale is used to adjust scale
  • The midsagittal vocal tract outline is traced on a new transparent layer
  • The tracing is saved with transparent background

Identifying articulator movements

  • Consecutive tracings are superimposed and compared in order to reveal local shifts in contours, reflecting movement.
  • This example shows tracings from five consecutive film frames, superimposed relative to the mandible.
  • This captures movements that can be attributed to activity in the muscles of the lower lip and the tongue, that are anchored to the mandible
  • In this example, the tongue blade was depressed away from the alveolarpalatal region (withdrawal from a vibrant [r]), while the tongue body was elevated (towards the hard palate) and advanced (expanding the pharyngeal region) for an oncoming [e]
  • The speaker did not move the lower lip during this particular sequence.

Combining individual profiles into animation sequences

  • Pictures 1-5 below are separations from the example above.
  • These are then organized into a GIF animation
  • On a web page, a GIF animation displays the pictures rapidly, creating an illusion of movement, like picture 6
  • The actual speed of presentation depends on the performance of your computer, and will usually be slower than the original speech. The original duration of this sequence was 67 msecs.
© Sidney Wood and SWPhonetics, 1994-2012