Rhythms of F0 and Formants in Sustained Vowels under Noises

Human phonations are controlled subtantially by the voices that the speakers hear from themselves, and this control contributes to the major part of audio-vocal feedback. Both clinical and laboratory studies demonstrate that this feedback is relatively automatic and is not easily suppressed or modified consciously. The underlying audio-vocal feedback system depends on the auditory inputs to sense the subtle changes in vocal fundamental frequency (F0) and the other acoustic energy such as formants, then the system make compensations for the alteration of auditory input before further utterance. As a consequence, in the subjects with hearing impairment, there are many patterns of disordered speech discovered clinically, such as elevated vocal F0, overuse of nasal voices.. (你可以加上粉多的聽障構音異常), etc. In our previous works, the audio-vocal feedback was evaluated using sustaining vocalization of vowel [a] and power spectral analysis of F0. The significant increases of low-frequency fluctuations of F0 were revealed both in the normal-hearings under noise masking and in the hearing impaired individuals. However, the spectral energy distributes so variously for different vowels, and the effects of audio-vocal feedback will be deduced differently for different vowels. The investigations of variability of F0 and formant shifts in different vowels may provide more information about the audio-vocal feedback and may also provide clues for the disordered speech in a hearing impaired subject.
With the growing understanding of the role of audio-vocal feedback in the control of F0, it’s important to explore this feedback throughout different vowels that are consisted of different articulatory characteristics and so as different acoustic features. Using the noises with different energy distribution, the acoustic energy of the vowels can be masked in different frequency bands. Therefore, the effects of audio-vocal feedback can be explored in individual frequency bands. Here, the healthy adults with gender evenly distributed and a age range between 20 and 40 years are planned to be enrolled. The voice samples to be collected include the sustained vowels [a], [i], and [m]. The vocalizations will be arranged randomly to avoid an order effect. A 2-min breaks is allowed between phonations to avoid the possible muscle memory effects of phonation. In addition, five interference conditions of auditory inputs will be introduced in the vocalizations of the vowels. The conditions are the following and are also arranged in a random order: 1) no masking (NO); 2) earphone only without noise (EO); 3) speech noise (SN); 4) high-pitch noise with equal energy over 1000 Hz (HN); 5) wideband noise with equal energy from 250 Hz to 8000 hz (WN). Each participant will be informed in detail about the upcoming changes of noises from the earphone, and then be asked to phonate as steady as possible within 70-80 dBA in modal register. Vocal intensity will be self-monitored by a real-time visual scale on the screen. By analyzing and comparing F0, perturbations of F0, oscillations of F0, and formant frequency between different noises and vowels, the mechanism of audio-vocal feedback can be investigated in individual frequency bands. Potentially, not only will it foster the application of hearing amplification and auditory rehabilitation, but also build a brand new vision of voice evaluation, and even the lesion detection of related neural loop.