VF_2006a - Vocal Aging in Male and Female Singers and...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Vocal Aging in Male and Female Singers and Nonsingers Singers James Harnsberger W.S. Brown, Jr. Howard Rothman Richard Morris Rahul Shrivastav 35th Annual Symposium: Care of the 35th Professional Voice Professional June 2, 2006 Acknowledgments Acknowledgments • Jeannette LoVetri • Douglas Hicks, Ph.D. Introduction Introduction • Questions of interest – What changes occur in voices with advancing What age? age? – How are these changes perceived? – Can a person’s perceived age be altered? • Professional training: Vocal training could increase vocal Professional fitness fitness • Theoretical relevance – Understanding speech disorders in the elderly Understanding population population – General models of aging Introduction Introduction • What physiological changes occur with vocal What aging? aging? – – – – Lengthening of the vocal tract or oral cavity Reduction in pulmonary function Reduction Laryngeal cartilage ossification Increased stiffening of vocal folds (particularly in Increased males) males) – Reduced closure of vocal folds (particularly in Reduced males) males) Introduction Introduction • Acoustic consequences with advancing Acoustic age? age? – Mean fundamental frequency (F0) Mean coalescence coalescence • Higher in older males • Lower in older females Introduction Introduction Vocal Aging: Mean F0 (Brown et al., 1991) 250 Mea n F0 (Hz) 200 192 175 150 127 118 Females (N = 20) Males (N = 15) 100 50 20-35 yrs 65-85 yrs Age Groups Introduction Introduction • Acoustic consequences with advancing age? – – – – F0 variability: increase Speaking rate: decrease Harmonics to noise ratio: increase Lowering of formant frequencies Introduction Introduction • How can we specify the perceptually relevant How cues to age in voice? cues – – Acoustic analysis Perceptual testing • Age estimation – Natural stimuli – Synthetic stimuli – Natural manipulated stimuli • Age difference estimation – Perceptual similarity evaluated using multidimensional scaling to extract number of perceptually relevant cues Introduction Introduction • Can perceived age be shifted through training? Can Evidence for: Evidence – Significant mismatch between perceived and actual age for Significant individual speakers individual – Health, independent of chronologic age, can influence Health, perceived age perceived – Anecdotal: Older trained professional singers may sound Anecdotal: younger younger “We might postulate that very fit older performers, who can add their vocal We performance, may not experience deleterious vocal changes as they get older.” (Boone, 1997 – Journal of Voice) older.” Purpose of study Purpose • Explore perceptually relevant cues to age in Explore male and female voices male – Age estimation task – Age difference estimation • Determine whether or not professional Determine training has efficacy in shifting the perceived age of male and female voices age – Age estimation task Experiment 1 Experiment • Stimulus Materials – Single sentence (second sentence from Single the Rainbow Passage) produced by: the • • • 20 Old Singers (10 M, 10 F) 20 Old Nonsingers (10 M, 10 F) 40 Foils – 10 Middle Aged Singers (5 M, 5 F) – 10 Middle Aged Nonsingers (5 M, 5 F) – 10 Young Singers (5 M, 5 F) – 10 Young Nonsingers (5 M, 5 F) Experiment 1 Experiment • Stimulus Materials – Singers and Nonsingers were equated for Singers chronological age chronological Female Male Singer 77.5 years 78.0 years Nonsinger 78.4 years 78.5 years Experiment 1 Experiment • Subjects and Procedures – 22 normal-hearing listeners – Two repetitions of the stimulus materials Two were presented free field in random order were – Subjects were instructed to estimate Subjects speaker age in years (no range provided). provided). Experiment 1 - Results Experiment • Experiment 1 - Age Estimation 70 Percei ved Age (Years) 65 Repeated Repeated Measures ANOVA ANOVA – 60 Singer 55 Nonsinger – – 50 45 40 Female Male Gender Group Training: F (1,21) Training: = 184, p < 0.01 184, Gender: n.s. Training*Gender: Training*Gender: n.s. Experiment 1 – Post hoc Analysis Experiment • Experiment 1 - Age Estimation (Foils) 50 – – 45 Percei ved Age (Years) Repeated Measures Repeated ANOVA ANOVA 40 Training: n.s All other factors and All interactions were significant significant Singer 35 Nonsinger • Post hoc tests – 30 25 – 20 Middle-Aged Female Middle-Aged Male Young Female Gender Group Young Male Young singers’ Young perceived age shifted towards middle-age towards Middle-aged singers’ Middle-aged perceived age was not affected by training affected Experiment 1 Results Experiment • Age estimation accuracy: Perceived vs. actual age – – – • Overall: r2 = 0.83** Nonsinger: r2 = 0.90** Singer: r2 = 0.77** Perceived age range narrower than chronological age Young voices overestimated, old voices underestimated Age Estimation 100 90 80 70 Actual Age – 60 Trained 50 Untrained 40 30 20 10 0 0 10 20 30 40 50 Perceived Age 60 70 80 90 Experiment 1 – Individual Speakers Experiment • Magnitude of difference in perceived age vs. Magnitude chronologic chronologic – Female Old Singers • – Male Old Singers • – 12 – 32 years younger (mean = 20) Female Nonsingers • – 10 – 26 years younger (mean = 20) 4 – 22 years younger (mean = 15) Male Nonsingers • 3 – 18 years younger (mean = 13) Discussion Discussion • Singer training shifted the perceived age of old and Singer young voices towards middle-age, regardless of gender. gender. • Speakers were accurately rank-ordered by age, but Speakers extremes of age range scale were avoided by listeners. listeners. • Question: What acoustic cues are most responsible Question: for perceived age in singers and nonsingers? for Acoustic Analysis Acoustic • Duration – – Old > Middle Aged, Young Singers = Nonsingers 4.5 4 4 3.5 3.5 G ender/Training G roup 5 4.5 Du ratio n ( seco nd s) 5 3 2.5 2 1.5 3 2.5 2 1.5 1 1 0.5 0.5 0 0 Young Middle Aged Age/Gender Group Old Old Female Singer Old Female Nonsinger Old Male Singer Duration (seconds) Old Male Nonsinger Acoustic Analysis Acoustic • Mean fundamental frequency – – Expected trend in male, but not female voices No significant trends for singer-nonsinger No distinction distinction 220 200 180 160 Female Male 140 120 100 Mean Fundam ental Frequency (Hz ) 240 220 Mean Fundamental Frequency (Hz ) 240 200 180 160 140 120 100 80 80 60 60 Young Middle Aged Age Group Old Old Female Singer Old Female Nonsinger Old Male Singer Gender/Training Group Old Male Nonsinger Acoustic Analysis Acoustic Fundamental frequency variability – Significant difference between singer and Significant nonsinger among female voices (not predicted) nonsinger 45 60 Fundam ental Frequency Standard Devi ati on (Hz ) F un damen tal Frequ ency St andard Deviatio n ( Hz ) • 40 35 30 25 20 15 10 5 0 Young Female Middle Age Female Old Female Young Male Age/Gender Group Middle Age Male Old Male 50 40 30 20 10 0 Old Female Singer Old Female Nonsinger Old Male Singer Age/Gender Group Old Male Nonsinger Future Work Future • Acoustic analysis of new cues – – • Voice quality Vowel formants Use larger talker samples – New Database • • • 50 Old (25 M, 25 F) 50 Middle-aged (25 M, 25 F) 50 Young (25 M, 25 F) ...
View Full Document

Ask a homework question - tutors are online