The sound generated while speaking is radiated from the mouth and nostrils, and also from the head and neck surfaces. As a result, Japanese scientists have used vibrometers to investigate speech by measuring vibrations brought on by phonation. By measuring vibrations, the scientists were able to analyze particular bands of frequencies.
Getting Set
The vibration patterns are seen on the surfaces of the head and neck, and how they influence the overall sound of speech, has not been extensively studied until now. A 22-year-old male with no speech disorder took part in laser vibrometry measurements, of which the aim was to image the patterns of vibration that are generated by speech.
The velocity of the vibrations was obtained by using a scanning laser Doppler vibrometer system (Polytec PSV-400-M4). This laser Doppler vibrometer is an optical transducer that can detect the shift in frequency of light reflected from a vibrating surface, caused by the Doppler effect. It can detect the velocity of vibrations as well as determining displacement at a certain point. The scanning vibrometer is also able to automatically scan and probe numerous points of a vibrating surface.
Figure 1 demonstrates the experimental setup. The vibrometer’s scanning head was mounted on a tripod and positioned perpendicular to the floor. The participant then lay directly beneath the scanning head.
The study participant repeatedly spoke, keeping his head completely still while measurements were being taken. A microphone recorded the sounds of his speech.
Measurements of the vibration patterns on the surface of the face were measured from the frontal direction, perpendicular to the forehead, as well as being measured from an oblique direction nearly perpendicular to the left cheek and the left side of the nose.
Firstly, system control software determined scanning points on the surface of the face, and then, during the measurement, the vibrometer scanned each point and detected the velocity of the vibrations. Each point took approximately one second to probe, and the vibration velocity and speech sounds were measured up to 5 kHz.
Results
The two upper images in Figure 2 show the vibration patterns of the frontal facial surface during sustained phonations.
There were notable differences between the vibration velocity patterns for the vowel (left) and nasal (right) consonant. For the vowel-consonant, the area of the facial surface around the mouth vibrated most strongly when compared to the other regions. Conversely, for the nasal consonant, the surfaces of the nose and the neighboring areas vibrated strongly, due to resonances in the nasal sinuses. Vibrations were also seen in the forehead to some extent, which could indicate that the frontal sinuses resonated during the sounding of nasal consonants.
The two lower images in Figure 2 show the vibration velocity patterns detected on the left facial surface for the phonemes. The vibration detected on the side of the nose was seen to be stronger for both phonemes than the vibrations indicated in the upper images of Figure 2. This result shows that the laser light’s direction plays a significant part in this method of measurement. This particular result also revealed that the nose surface vibrated even when study participant sounded nasal and vowel consonants.
.jpg)
Figure 1. Experimental setup.
![Vibration velocity patterns of frontal and left facial surface during articulation of vowel (left) and nasal (right) consonant. The unit is m/s [dB] and 0 dB is equal to 1 m/s.](https://d36nqgmw98q4v5.cloudfront.net/images/Article_Images/ImageForArticle_1598(2).jpg)
Figure 2. Vibration velocity patterns of frontal and left facial surface during articulation of vowel (left) and nasal (right) consonant. The unit is m/s [dB] and 0 dB is equal to 1 m/s.
How Can It Be Used?
The method described in this study allows us to assess speech in patients with a cleft palate or velopharyngeal insufficiency. Being able to determine the pattern of vibration could be beneficial in the form of visual feedback from a speaking exercise, for instance. The vibration pattern may be easier to relate to their somesthesis than spectra of their speech sounds. This method of measurement could also be used in singing exercises.
Conclusion
The method described in this paper facilitates fast, non-contact and multi-point measurements of the vibration velocity of skin surfaces, and the results of such measurements will expand our knowledge of speech production.
The next step in the research is to investigate the connections between the vibration velocity pattern of skin surfaces and the formants and antiformants of speech sounds.

This information has been sourced, reviewed and adapted from materials provided by Polytec.
For more information on this source, please visit Polytec.