Abstract - Speech Production

Voice Characteristics of Younger and Older Speakers

Pascal van Lieshout1,2,3,4, Dave Fernandes2,5, Kathy Pichora-Fuller1,3,
Jessica Banh1, Konstantin Naumenko1, Huiwen Goy1
1 Department of Psychology, University of Toronto, Mississauga, Canada
2 Department of Speech-Language Pathology, Oral Dynamics Lab, University of Toronto, Toronto, Canada
3 Toronto Rehabilitation Institute, Toronto, Canada
4 Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, Canada
5 Hospital for Sick Children, Toronto, Canada

 Changes in voice quality serve as indicators of various pathologies, including structural disorders of the voice production system such as vocal nodules, functional problems with the voice due to work-related stress or fatigue, and neurological disorders such as Parkinson’s disease. Knowing the range of values typically found in a healthy population provides a baseline for disease detection and for monitoring a patient’s progress in therapy. The goal of this project was to collect a large number of voice samples from healthy younger and older adults, to map the distribution of values for various voice quality measures in a healthy population. Participants in this study were asked to say the vowel /a/ in different ways: using their regular voice, using the highest pitch they could reach, for as long as possible, and as softly as possible. Participants also provided examples of their speech by reading a paragraph from the Rainbow Passage (Fairbanks, 1960). Examples of voice measures taken in this study include fundamental frequency (pitch), maximum phonation time (an indicator of airflow control), shimmer and jitter (measures of variability in the voice which correlate with hoarseness), and noise-to-harmonics ratio (which correlates with breathiness).

Results  We collected a wide variety of measures from the speech and voice samples of healthy younger and older adults, including some measures that once lacked normative data. We found that men and women’s voices do not change in the same way with age; for instance, older women in our study had lower-pitched voices than younger women, but younger and older men had similar pitches overall. After measuring a much larger sample of people than in past studies and carefully controlling the recording conditions, we found that age-related changes in the voice are not as large as some past studies have indicated. Descriptions of group results from this study have recently been published in the Journal of Voice.

Back to Speech Production


                                                                     

                                      Speech and Voice Rating Study

                    Huiwen Goy, Kathy Pichora-Fuller, Pascal van Lieshout

         The quality of communication is affected by listeners’ perception of talkers’ characteristics. The purpose of this study was to find out which acoustic characteristics of normal voices lead everyday listeners to judge voices as being “good” or “poor”, and whether the voices of older adults are perceived to be better or worse than the voices of younger adults. Both younger and older adult listeners participated in this study. They heard short sentences and samples of the vowel [a] spoken by male and female speakers of different ages. After listening to each voice, participants were asked to decide on the speaker’s gender and age, and to rate the speaker on qualities such as pleasantness and clarity of articulation for speech samples and qualities such as roughness for vowel samples.

         The results showed that listeners are near-perfect at identifying whether a speaker is male or female from hearing a short sentence, but that older listeners are more accurate than younger listeners at identifying the speaker’s gender from a simple vowel. When we modelled the relationships between listeners’ ratings and the acoustic characteristics of voices, these models suggested that older adults make use of more of the available acoustic information compared to younger listeners to judge how old a speaker is. The models also showed that listeners perceived speakers to be younger if they spoke with a faster rate and with a higher pitch. However, these models did not show which particular characteristics of speech led listeners to judge speech as being more pleasant or clear. Older listeners did not show any particular preference for younger or older voices, but younger listeners found older speakers’ vowels less pleasant and less smooth than younger speakers’ vowels.

         Younger and older talkers were judged to be similar on most qualities except age; listeners did perceive the voices of older adults as belonging to speakers who were older, but they also underestimated older speakers’ true ages by several decades. This result may stem from listeners having stereotypes about older voices that do not reflect the true voice characteristics of most older adults. Alternatively, vocal aging may happen at a different rate than chronological aging, leading to difficulties in judging a speaker’s chronological age from the sound of their voice.

         In sum, the acoustic characteristics of voices affect listeners’ perception of how old a speaker is, but it is not yet clear which acoustic characteristics of voices lead listeners to judge speakers as pleasant or clear. However, the results from this study suggest that the perception of speaker quality may be more affected by the age of the listener than by the age of the talker.

Back to Speech Production