Effect of Hydration and Vocal Rest on the Vocal Fatigue in Amateur Karaoke Singers

Effect of Hydration and Vocal Rest on the Vocal Fatigue in Amateur Karaoke Singers Edwin M-L Yiu and Rainy MM Chan Hong Kong, China Summary: Karaoke singing is a very popular entertainment among young people in Asia. It is a leisure singing activity with the singer s voice amplified with special acoustic effects in the backdrop of music. Music video and song captions are shown on television screen to remind the singers during singing. It is not uncommon to find participants singing continuously for four to five hours each time. As most of the karaoke singers have no formal training in singing, these amateur singers are more vulnerable to developing voice problems under these intensive singing activities. This study reports the performance of 20 young amateur singers (10 males and 10 females, aged between 20 25 years) on a series of phonatory function tasks carried out during continuous karaoke singing. Half of the singers were given water to drink and short duration of vocal rests at regular intervals during singing and the other half sang continuously without taking any water or rest. The subjects who were given hydration and vocal rests sang significantly longer than those who did not take any water or rest. The voice quality, as measured by perceptual and acoustic measures, and vocal function, as measured by phonetogram, did not show any significant changes during singing in the subjects who were given water and rest during the singing. However, subjects who sang continuously without drinking water and taking rests showed significant changes in the jitter measure and the highest pitch they could produce during singing. These results suggest that hydration and vocal rests are useful strategies to preserve voice function and quality during karaoke singing. This information is useful educational information for karaoke singers. Key Words: Karaoke singing Hydration Vocal rest Acoustic analysis Perceptual evaluation Phonetogram. Accepted for publication September 16, 2002. From the Voice Research Laboratory, Division of Speech and Hearing Sciences, The University of Hong Kong, Hong Kong, China. Presented at the 30th Annual Symposium of the Voice Foundation, Philadelphia, Pennsylvania, June 2001. Address correspondence and reprint requests to Edwin Yiu, PhD, Voice Research Laboratory, Division of Speech and Hearing Sciences, The University of Hong Kong, 5/F Prince Philip Dental Hospital, 34 Hospital Road, Hong Kong. E-mail: eyiu@hku.hk Journal of Voice, Vol. 17, No. 2, pp. 216 227 2003 The Voice Foundation 0892-1997/2003 $30.00 0 doi:10.1016/s0892-1997(03)00038-9 216

VOCAL FATIGUE IN AMATEUR SINGERS 217 INTRODUCTION Karaoke singing is a popular entertainment activity for many young people in Southeast Asia, especially in Hong Kong. It is also gaining popularity in North America and Europe. Karaoke singing is a leisure singing activity with the provision of background music through an amplifier. Music video, with song captions, is shown on a television monitor screen. The singer sings into a microphone and the voice is mixed with the background music and some special effects, such as echo, via an amplifying system before output to the speakers. Karaoke singers frequently sing continuously for four or five hours among loud background noise where people talk. This continuous use of voice in a wide pitch range and a high intensity level puts the karaoke singers at risk of developing dysphonia. 1 Most karaoke singers are amateurs; very few of them have formal singing training. Untrained singers usually strain their voice more often than trained singers in order to produce a wide range of frequency. 2 Gelfer, Andrew and Schmidt 3 found untrained singers to demonstrate poorer voice quality than trained singers following one hour of reading aloud. Therefore, a lack of formal singing training in these amateur karaoke singers may make them more susceptible to developing voice problems after intensive singing. In our experience (Yiu s personal observation, 2002), it is not uncommon to find karaoke singers complaining of vocal fatigue following singing. Vocal fatigue usually refers to tiredness of voice following prolonged speaking. 4,5 It requires an individual to use more effort to continue speaking 6 and may be accompanied by changes in vocal quality, loudness, pitch, effort in voicing, 7 and feeling of laryngeal discomfort such as laryngeal aching, throat fullness, soreness, neck tightness, and pharyngeal/ laryngeal dryness. 5 Singers often relate vocal fatigue to a reduced ability to project or to sustain voice, a reduced power of voice, a reduced pitch and loudness range, an increased hoarseness level, an increased effort to produce voice, or a general vocal constriction. 8 Vocal fatigue usually occurs following the use of abnormal pitch, abnormal intensity, 4 abnormal voice quality, 1 or continuous reading for over an hour. 3,5 It has been reported to be associated with anxiety, 7 changes in weather, lack of sleep, and increased physical activity. 9 Changes associated with vocal fatigue It is difficult to quantify vocal fatigue directly. Physiological changes associated with vocal fatigue, such as voice quality or vocal function, are often taken as the measures to reflect vocal fatigue. The measurements reported in the literature include perceptual quality rating, acoustic measures, aerodynamic measures, videostroboscopic examination, and phonetogram (voice range profile) study. Perceptual voice quality changes Some reports on vocal fatigue focused on investigating perceptual voice quality. Breathy and strained voice qualities have been reported in teachers who complained of vocal fatigue. 7 Abnormal voice pitch has also been reported as a sign of vocal fatigue in professional voice users. 10 Acoustic voice quality changes Acoustic measures have also been used to study vocal fatigue. Changes in fundamental frequency of sustained vowel and reading have been noted following prolonged voice use. 3,5,11 The change in fundamental frequency may be gender-dependent. In a study of a group of trained theatre performers, Novak, Dlouha, Capkova and Vohradnik 11 noted an increase in the fundamental frequency in men but a lowering of fundamental frequency in women following their performances. In other reports, it has been found that an hour-long reading could lead to a significant increase in fundamental frequency, 5 and a reduction in the signal-to-noise ratio. 3 Aerodynamic changes In the study by Stemple, Stanley and Lee, 5 they found no significant changes in the phonation volume, airflow rate, or the maximum phonation time following prolonged reading. In a later report, Eustace, Stemple and Lee 6 found a significantly shorter phonation time and higher airflow rate in another group of subjects following prolonged voice use. In a study by Kostyk and Rochet, 12 a different finding was reported. They found a reduced airflow rate following vocal fatigue. The contradictory findings

218 EDWIN YIU AND RAINY CHAN between that of Eustace et al 6 and Kostyk and Rochet 12 might have been caused by the use of different strategies by the individuals in response to vocal fatigue. One group might have used more respiratory drive than the other in trying to compensate for the vocal fatigue. Therefore, the airflow rate measures might have been an indication of the compensatory strategy to vocal fatigue rather than a consequence. Other aerodynamic measures have also been used to study vocal fatigue. For example, Solomon and DiMattia 13 found the phonation threshold pressure to increase significantly following a two-hour long reading. Vilkman, Lauri, Alku, Sala and Sihvo 14 found the subglottal pressure and inverse filter glottal flow waveform to increase following prolonged reading. However, as noted earlier with the airflow measure, these aerodynamic measures may also reflect a respiratory and laryngeal compensatory effort in response to the vocal fatigue. 14 Therefore, aerodynamic measurements may not be appropriate measures for vocal fatigue. Videostroboscopic changes Videostroboscopic data are generally qualitative and subjective in nature. They, nevertheless, provide useful information about vocal function. Scherer, Titze and Raphael 15 and Mann et al 16 found an increase in vocal fold edema in subjects who read aloud for more than an hour. Stemple and his colleagues 5,6 reported development of anterior glottal chinks and incomplete or spindle-shaped closure of the glottis in individuals who read aloud for two hours. However, Stemple and his colleagues 5,6 found no changes in the amplitude of the vocal fold movement, mucosal wave, phase closure, and phase asymmetry. On the contrary, Mann et al 16 found a decrease in vocal fold mucosal wave and amplitude. Phonetogram changes Phonetogram, or voice range profile, measures the fundamental frequency range and the loudness range that a person can achieve. 17 The voice range is usually displayed with the loudness (db SPL) as the vertical axis and the fundamental frequency range (Hz) as the horizontal axis. The use of phonetogram in clinical voice assessment has attracted much interest in the last decade. 17 However, very few reported studies on vocal fatigue make use of phonetogram, with the exception of that by Van Mersbergen, Verdolini and Titze. 18 They studied the changes in voice range profile in 10 vocally-untrained and healthy females, aged between 18 to 35, following a day of voice use and found no significant changes in the voice range profile between the morning and evening recordings. Although this study showed a negative result, more studies are needed to investigate fully the potential of phonetogram in studying vocal fatigue. The studies reviewed above investigated vocal fatigue by inducing prolonged voice use experimentally. Some of the prolonged voice use tasks employed in these studies were of a set duration. For example, one study asked the subjects to read aloud for one hour, 3 and another for two hours. 5 Others assumed vocal fatigue to occur after a day of voice use and investigated whether there were any changes at the end of a day. 18 These studies assumed vocal fatigue would occur after some extensive voice use. However, the amount of voice use that leads to vocal fatigue may vary from individual to individual. There is a possibility that some of the subjects reported in these studies might not have developed vocal fatigue at the end of these set tasks. To avoid such possibility, the present study did not fix the amount of time for the subjects to perform the vocal task. Instead, a singing task was chosen and the subjects were encouraged to sing for as long as they could until they felt tired. The present study aimed to determine the amount of singing that would lead to the perception of vocal fatigue in 20 untrained amateur singers. This study also aimed to determine whether regular hydration and brief vocal rests during singing would minimize the changes in vocal quality and function as measured by perceptual voice evaluation, acoustic analysis, voice range profile analysis, and subjects self-perception of vocal fatigue. It was hypothesized that a combination of hydration and vocal rests during continuous karaoke singing would prolong normal vocal function and preserve the voice quality. METHODS Self-perception of vocal tiredness was taken as the working definition of vocal fatigue in this study. Vocal fatigue was defined as the feeling of changes

VOCAL FATIGUE IN AMATEUR SINGERS 219 in the ability to project or to sustain voice. In daily life, vocal fatigue may involve pitch, loudness, and quality change and is usually associated with throat discomfort, pain, or dryness. Participants Ten Hong Kong Chinese males and 10 females participated in the study. Each male subject was matched in age with a female subject within three years of age (mean 21.5 years, standard deviation 0.85, range 20 to 23 years). Fourteen (70%) of them were students, four (20%) worked as clerical staff, and two (10%) were engineers. All the participants also satisfied the following selection criteria: 1. Reported to have normal vocal quality. 2. Required only to talk occasionally in daily work. 3. Reported to experience vocal fatigue after karaoke singing. 4. Participated in regular karaoke singing for at least two days a week. 5. Were nonprofessional singers. 6. Had no formal voice or singing training. 7. Had no chronic medical disorders, such as thyroid, respiratory, or psychiatric problems. 8. Had no history of voice problems. 9. Were not on any regular medication which might have an effect on voice quality. 10. Had no respiratory tract infection on the day of assessment. 11. Were nonsmokers. 12. Were not recreational drug users. 13. Were not alcohol drinkers (consumed less than one glass of beer per week). 14. Had normal voice on the day of assessment as determined by the second author. To control for the possible hormonal effects on voice quality, the female participants were not menstruating and not taking contraceptive pills at the time of the study. Procedures Karaoke singing task Each participant was asked to sing in a quiet room with karaoke facility (Panasonic SL-VP35), which provided music video on a television and background music with echo effects. The participants sang only the songs with melodies that suited their particular gender pitch range. The songs were presented in the same predetermined sequence for each gender group. The volume of the background music and echo effect were set at the same level for all participants. The participants were required to sing continuously until they reported feeling fatigue with their voices and could not sing anymore. Five male and five female subjects were randomly assigned to a group that was given hydration and voice rests (HVR) during singing. Each participant was givenaone-minute vocal restand100 ml ofwater after singing each song. The other 10 subjects (five males and five females) were not given hydration or voice rests (nonhvr) during singing. Voice recordings for acoustic and perceptual analysis A baseline recording (R-1) was carried out just before the singing task for each subject. There were three post singing recordings, one carried out after singing 10 songs (R-2). The second one was carried out after singing additional five songs (R-3), and the last recording (R-4) was carried out after the last song when the participant reported vocal fatigue and could not sing anymore. Each recording required the subject to sustain an /a/ phonation (for five seconds) and to read aloud a Chinese sentence /ba ba da bɔ/(father hits the ball) at a comfortable pitch and loudness level. The sentence was chosen because all the words were made up of nonaspirated consonants and simple vowels. Each vowel and sentence was recorded three times. All recordings were carried out using Kay Elemetrics Computerized Speech Lab (CSL, Key Elemetrics, Lincoln Park, NJ) with a Shure SM48 microphone at a mouth-tomicrophone distance of 10 cm. After each recording session was completed, each participant listened to all recorded samples (three vowels and three sentences) together with the second author. A vowel and a sentence, which both of them agreed to demonstrate the best quality, were selected to represent the best recordings of that session. These two selected samples were used for subsequent acoustic and perceptual voice analysis. The selection of only the best recordings in each session was intended to detect subtle quality deteriorations in the voices

220 EDWIN YIU AND RAINY CHAN which the subjects could best produce. The investigators believed that if voice quality deteriorations could be found even among the best produced voices, it would be an indication of vocal fatigue that was beyond the control of the subjects. Voice recordings for phonetogram analysis Phonetogram recordings were carried out on three occasions only (R-1, R-2, and R-4). As each recording took approximately 45 minutes, only the baseline recording (R-1) and two recordings following singing were undertaken to reduce the time for the overall procedure and also avoid distracting the subjects with the recording tasks. The phonetogram recordings were carried out using the Tiger Electronic Dr. Speech Phonetogram and a Shure SM48 microphone. The mouth-to-microphone distance was kept at a fixed distance of 10 cm. In each recording, the participants were asked to produce low and high pitches at different loudness levels. The male participants sustained the vowel /a/ that matched the tone generated at musical note C 3 (131 Hz), while the female participants sustained the vowel /a/ that matched the tone generated at note C 4 (262 Hz). Each participant produced this initial pitch first at a comfortable loudness level, and then gradually reduced the loudness to the softest level by gliding down the intensity level. The subject then produced higher pitches through the musical scale. Each pitch was produced with reducing loudness from the subject s most comfortable loudness until the softest voice without whisper was reached for each pitch. The subject then went down the musical scale from the initial C 3 or C 4 note with decreasing pitch. Each pitch was produced with decreasing loudness. The recording procedures were repeated again with all the pitches using an increasing loudness with each pitch. This phonetogram recording task took the subjects away from singing for approximately 45 minutes. Although the demand of the phonetogram recording was not identical to the singing task, the subjects were still required to use their voice continuously and strenuously during the recording. Data analysis Duration of singing before feeling fatigue The total amount of singing time was recorded for each subject. The resting time between every two songs for the HVR group and the time taken for the instrumental recordings were excluded from this record. Acoustic analysis Mean fundamental frequency, jitter percent, shimmer db, and noise to harmonic ratio were obtained for the selected /a/ and sentence samples from each recording using Kay Electmetrics Computerized Speech Lab (CSL 4300B) and the Multidimensional Voice Program (MDVP 4305). For the analysis of vowels, a middle three-second portion of each vowel was used. For the analysis of the sentences, the onset of the first consonant and the offset of the lastvowel of each sentence production were segmented to include the sentence for analysis. The MDVP option for analyzing connected speech was used in the sentences analysis as it has been shown by Yiu et al 19 that this option produced reliable acoustic analysis for connected speech. Perceptual analysis Three final year speech pathology students with a year of clinical experience with voice patients were recruited to rate the roughness and breathiness of each of the sentences /ba ba da bɔ/. Roughness and breathiness were chosen for evaluation, as the ratings for these two vocal qualities have been reported in the literature to be more reliable than for the other voice qualities 20,21. In this study, roughness was defined as audible irregular or uneven quality with a lack of clarity. 22,23 Breathiness was defined as audible air emission, sound of expiration, or friction noise during phonation. 22,23 The three judges were given a perceptual training program using a stimulus response stimulus feedback paradigm. The program involved listening and rating 10 male and 10 female voice stimuli with varying degrees of roughness and breathiness severity. The training program took approximately an hour to complete and has been shown to improve the reliability of perceptual voice evaluation using anchors. 24. The same sentence stimuli that were used in the acoustic analysis were also used for the perceptual rating. The stimuli were separated into male and female sets, each with 40 stimuli (four recordings for each subject). Eight stimuli from each gender set were randomly selected and repeated once, therefore

VOCAL FATIGUE IN AMATEUR SINGERS 221 resulting in a total of 48 stimuli in each gender set of voice stimuli. These repetitions, which constituted 20% of all stimuli, were used to determine the intrarater reliability in the perceptual rating. All the judges rated the male voice stimuli first. The stimuli were presented in a random order. The female voice stimuli were rated on another day. Natural voice references or anchors representing just noticeable and severe roughness and breathiness were provided for each gender set to assist the judges during the rating process. The judges were allowed to listen to each voice stimulus and anchor for as many times as necessary. Each judge was required to give a roughness and a breathiness rating for each stimulus using two separate 10-cm long scroll bars on a 38-cm wide computer screen. The right end of the scroll bar was labeled as severe and the left end was labeled as just noticeable. The ratings made by the scroll bars were measured to the nearest 1 mm. The ratings givenby the three judgesfor each voice stimulus were averaged to give a final rating for each stimulus. Phonetogram analysis The phonetogram provided measurements for the range of fundamental frequency (Hz) and loudness (db). These were derived from the highest and lowest fundamental frequency (Hz) and also the maximum and minimum loudness (db) that an individual could produce. The total area under the voice range (db semitone) was also obtained from the phonetogram analysis. RESULTS Nonparametric statistics were used for the analysis, as the sample size was small. As the data set was small in each group (five subjects in each gender group), probability was therefore calculated for each test using the exact distribution of the test statistics. 25 The calculations were carried out using the SPSS Exact Tests. 25 Amount of singing time before vocal fatigue The mean durations of singing before the subjects reporting vocal fatigue are listed in Table 1. There was no significant difference between the male and female subjects within the hydration and vocal rest TABLE 1. Amount of singing duration before feeling fatigue Groups Gender Mean Standard Mini- Maxi- (minute) Deviation mum mum HVR Non HVR Male 101.74 3.81 96.47 105.47 Female 102.12 3.50 97.02 105.43 Male 86.15 3.79 82.53 91.60 Female 84.80 5.45 78.02 92.40 HVR group with Hydration (100 ml water) and Vocal Rest (one-minute) given between every song; NonHVR group non-hydration and non-vocal Rest group (HVR) group (Mann Whitney U 12, Z 0.11, p 0.97) or within the nonhydration and nonvocal rest (nonhvr) group (Mann Whitney U 12, Z 0.10, p 0.98). Therefore, the two gender groups were combined for further analysis. The HVR group sang on average more than 100 minutes. This was significantly longer than the non-hvr group, which sang less than 90 minutes on average (Mann Whitney U 0, Z 3.79, p 0.001). In addition, the shortest singing time shown by the HVR group (96.47 minutes) was still longer than the longest singing time in the nonhvr group (92.40 minutes). Acoustic analysis Table 2 shows the mean acoustic values of the /a/ in the HVR and nonhvr groups across the four recording sessions. Nonparametric Friedman tests with exact probabilities calculated were carried out for each acoustic measure across the four recordings in each of the gender and experimental group. No significant results were found for any of the acoustic measures in any of the groups (all p 0.1, see Table 2). Table 3 presents the mean acoustic values of the sentence /ba ba da bɔ/ produced by the HVR and nonhvr groups across the four recordings. A significant change in the mean jitter percent was found in the male nonhvr groups (c 2 9.24, p 0.02; see Table 3). Further planned Wilcoxon signed ranked tests showed a significant higher jitter in R- 2 when compared with those in R-1, R-3, and R-4 (all Z 2.02, p 0.03).

222 EDWIN YIU AND RAINY CHAN TABLE 2. Means acoustic measures of /a/ across the four recordings Recordings Friedman tests Groups Gender R-1 R-2 R-3 R-4 c 2 p Mean fundamental frequency (Hz) HVR Male 141.37 146.17 145.49 149.56 2.52 0.47 Female 254.43 261.45 244.83 246.55 4.20 2.04 NonHVR Male 148.69 140.60 134.95 131.49 2.28 0.52 Female 248.95 251.97 244.95 255.69 3.00 0.39 Mean jitter (%) HVR Male 0.73 0.66 0.53 0.57 7.52 0.47 Female 1.34 1.26 1.37 1.23 2.54 0.47 NonHVR Male 0.43 0.44 0.85 0.67 3.48 0.32 Female 1.47 1.38 1.57 1.59 1.08 0.78 Mean shimmer (db) HVR Male 0.26 0.26 0.26 0.24 3.37 0.34 Female 0.33 0.29 0.36 0.35 4.44 0.22 NonHVR Male 0.23 0.26 0.31 0.38 0.73 0.86 Female 0.30 0.34 0.31 0.29 0.60 0.89 Mean noise-to-harmonic ratio HVR Male 0.14 0.13 0.12 0.13 1.89 0.59 Female 0.13 0.12 0.12 0.14 1.08 0.78 NonHVR Male 0.13 0.13 0.16 0.16 6.32 0.10 Female 0.13 0.12 0.13 0.13 0.43 0.93 R-1 presinging recording; R-2 recording after 10 songs; R-3 recording after 15 songs; R-4 recording after subject reported vocal fatigue HVR group with Hydration (100 ml water) and Vocal Rest (one-minute) given between every song; NonHVR group non- Hydration and non-vocal Rest group Perceptual voice evaluation Reliability and agreement measures Mean Pearson s correlation coefficients for intrarater reliability were 0.82 and 0.89 for the ratings of male and female voices respectively (p 0.001). Mean Pearson s correlation coefficients for interrater reliability among the three listeners were 0.75 and 0.85 for the male and female voices, respectively (p 0.001). Interrater agreements were calculated to determine how well two raters agreed within 1 cm on the rating scale in rating each stimulus. The mean agreements among the three raters were 0.86 and 0.82 for the male and female voice, respectively. Mean roughness and breathiness ratings Table 4 presents the group mean perceptual ratings for the rough and breathy qualities. It should be noted that the mean perceptual ratings for both qualities were close to normal or with very low severity ratings (smaller than 2 ). Friedman tests with exact probabilities calculated were performed to determine if there were significant changes in the severity level of perceptual qualities over the four recordings. No significant changes were noticed in any of the perceptual ratings (p 0.2, see Table 4). Kruskal- Wallis tests also showed no significant results in the breathiness and roughness ratings in any of the four recordings among the two gender groups in the HVR and nonhvr groups (p 0.05). Phonetogram analysis Table 5 lists the mean results of seven phonetogram measures for the HVR and nonhvr groups. These measures include mean fundamental frequency range, mean highest fundamental frequency, mean lowest fundamental frequency, mean loudness range, mean maximum loudness, mean minimum loudness, and mean total area of voice range profile.

VOCAL FATIGUE IN AMATEUR SINGERS 223 TABLE 3. Means acoustic measures of /ba ba da bɔ/ across the four recordings Recordings Friedman tests Groups Gender R-1 R-2 R-3 R-4 c 2 p Mean fundamental frequency (Hz) HVR Male 130.15 134.03 130.19 130.25 0.60 0.89 Female 244.30 238.84 240.10 242.18 2.04 0.56 NonHVR Male 134.08 167.94 141.60 138.28 1.32 0.72 Female 238.70 239.27 226.95 245.66 2.04 0.56 Mean jitter (%) HVR Male 2.13 1.90 2.02 2.01 0.60 0.89 Female 2.85 2.56 2.57 2.27 0.60 0.89 NonHVR Male 1.32 1.92 1.21 1.12 9.24 *0.02 Female 2.47 2.39 2.88 2.44 0.60 0.89 Mean shimmer (db) HVR Male 0.93 1.13 0.90 0.77 4.20 0.24 Female 0.61 0.67 0.62 0.62 1.08 0.78 NonHVR Male 0.72 0.56 0.63 0.63 4.20 0.24 Female 0.62 0.67 0.63 0.56 1.16 0.76 Mean noise-to-harmonic ratio HVR Male 0.31 0.24 0.24 0.24 1.30 0.72 Female 0.32 0.28 0.23 0.21 7.78 0.05 NonHVR Male 0.26 0.26 0.20 0.21 6.84 0.08 Female 0.32 0.27 0.27 0.23 1.56 0.67 R-1 pre-singing recording; R-2 recording after 10 songs; R-3 = recording after 15 songs; R-4 recording after subject reported vocal fatigue HVR group with Hydration (100 ml water) and Vocal Rest (one-minute) given between every song; NonHVR group non- Hydration and non-vocal Rest group *Significant at 0.05 level The HVR group showed no significant changes in any of the measures over time (p 0.1, see Table 5). With the nonhvr subjects, a significant reduction in the highest fundamental frequency was noticed in the female groups (c 2 6.71, p 0.03). Wilcoxon signed-ranked tests were then performed between the three recordings to determine where the difference existed. It was shown that the mean highest fundamental frequency at R-2 (610.2 Hz) was significantly lower than that of R-1 (817.4 Hz; Z 2.02, p 0.04). There was no significant difference between R-4 and R-2 or between R-4 and R-1 (p 0.1). DISCUSSION An objective of this study was to determine the amount of singing that would lead to the feeling of vocal fatigue and changes in vocal quality and function in untrained amateur singers. The group that was given water and vocal rests (HVR group) sang significantly longer (mean 101.93 minutes) than the group without taking water or rests (nonhvr group, mean 85.48 minutes). This result suggests that frequent brief hydration and vocal rests were useful in prolonging singing time before feeling vocal fatigue during karaoke singing in amateur singers. The second objective was to determine whether frequent brief vocal rests and hydration between songs would affect the changes in vocal quality and function following intensive karaoke singing. The HVR group in the present study showed no significant changes in any of the acoustic measures, perceptual or phonetogram measures. The nonhvr male subjects showed a significant increase in jitter percent after singing 10 songs (R-2). This change in

224 EDWIN YIU AND RAINY CHAN TABLE 4. Mean perceptual ratings of /ba ba da bɔ/ across the four recordings Recordings Friedman tests Groups Gender R-1 R-2 R-3 R-4 c 2 p Mean Roughness HVR Male 1.81 1.42 1.35 1.24 3.73 0.29 Female 0.65 0.71 0.60 0.52 1.53 0.67 NonHVR Male 1.37 1.33 1.49 1.61 3.24 0.36 Female 1.01 1.72 1.97 1.64 2.04 0.56 Mean breathiness HVR Male 0.27 0.39 0.22 0.43 1.17 0.76 Female 0.63 0.71 0.73 0.45 1.32 0.72 NonHVR Male 0.21 0.33 0.18 0.30 2.07 0.56 Female 0.62 1.08 1.25 1.21 4.20 0.24 R-1 = pre-singing recording; R-2 recording after 10 songs; R-3 recording after 15 songs; R-4 recording after subject reported vocal fatigue HVR group with Hydration (100 ml water) and Vocal Rest (one-minute) given between every song; NonHVR group non- Hydration and non-vocal Rest group jitter is also supported by similar findings from previous studies which showed changes in jitter following vocal fatigue. 3,5 The jitter measure seemed to return to normal on subsequent recordings (R-3 and R-4). This might have reflected a warm-up effect in which an initial deterioration in jitter occurred after singing 10 songs but the vocal function adjusted to the intensive singing demand subsequently. Nevertheless, this finding should be interpreted with caution, as this is the only acoustic measure which showed such a change. Neither the HVR nor nonhvr groups showed any changes in the perceptual roughness and breathiness ratings over time. It was likely that the perceptual procedure was not sensitive enough to detect the minimal changes as the mean ratings were all very mild (less than 2 out of the 10-point scale). There was, however, one additional change found when the phonetogram was used. The female subjects in the nonhvr group showed a significant reduction in the highest fundamental frequency that the subjects could produce. We postulate this reduction in vocal performance, as a result of the fatigue of the laryngeal muscles, was probably caused by the changes in the elasticity, the length, and the size of the muscles. Contrary to the expectation that there would be changes in the vocal quality and function when the subject reported vocal fatigue, this study found no detectable change in R-4 when compared to the other earlier recordings. The fact that there was no detectable change in the breathiness and roughness ratings suggests these two perceptual qualities were not appropriate measures to detect vocal fatigue in subjects with normal voice following intensive singing. Jitter measure has been shown by a number of investigators, like Titze 26 and Yiu, 27 to be a valid measure for analyzing voice signals which are periodic in nature only. Although the significant increase in the sentence jitter after singing 10 songs in the male nonhvr group might have been a warmup effect as discussed earlier, the sensitivity of the jitter measure suggest that jitter may be useful to reveal the effect of vocal fatigue. The highest sustainable fundamental frequency also demonstrated significant change in the female nonhvr group after singing 10 songs. This also appears to be a useful measure to study the effect of vocal fatigue on voice quality. However, more evidence is needed before a definite conclusion can be drawn. In summary, this study showed some evidence that hydration and vocal rest reduced the effect of vocal fatigue. The subjects that were given hydration and vocal rests were capable to sing longer. Both the HVR and nonhvr groups were able to maintain the vocal quality and function in general, as measured by perceptual and instrumental means. However, the nonhvr group demonstrated relatively more impairment.

VOCAL FATIGUE IN AMATEUR SINGERS 225 TABLE 5. Mean phonetogram (voice range profile) values across recordings Recordings Friedman tests Groups Gender R-1 R-2 R-4 c 2 p Mean fundamental frequency range (Hz) HVR Male 685.80 664.80 879.20 2.63 0.27 Female 838.40 715.60 723.80 0.40 0.82 NonHVR Male 845.20 849.20 741.60 1.44 0.49 Female 706.60 498.80 508.60 4.80 0.09 Men highest fundamental frequency (Hz) HVR Male 756.00 729.20 947.40 2.47 0.29 Female 962.80 844.60 808.60 1.41 0.49 NonHVR Male 910.00 907.20 802.00 1.44 0.49 Female 817.40 610.20 679.20 6.71 *0.03 Mean lowest fundamental frequency (Hz) HVR Male 70.00 64.60 67.40 0.44 0.80 Female 122.20 128.80 119.00 0.00 1.00 NonHVR Male 64.80 57.60 61.00 1.00 0.61 Female 112.96 108.16 136.60 2.84 0.24 Mean loudness range (db) HVR Male 45.70 42.40 52.92 0.40 0.82 Female 42.50 39.50 40.40 1.60 0.45 NonHVR Male 45.74 45.58 46.64 1.60 0.45 Female 41.46 38.78 38.10 1.60 0.45 Mean maximum loudness (db) HVR Male 116.54 118.58 120.00 2.00 0.37 Female 115.26 114.48 115.82 1.00 0.61 NonHVR Male 119.56 120.00 119.28 1.00 0.61 Female 111.82 108.60 109.72 0.00 1.00 Mean minimum loudness (db) HVR Male 70.84 76.18 73.36 1.60 0.45 Female 73.18 74.96 76.02 0.40 0.82 NonHVR Male 73.80 74.42 74.36 0.74 0.69 Female 70.30 69.76 71.56 1.60 0.45 Mean total area of voice range profile (db semitone) HVR Male 644.48 448.10 625.76 2.80 0.25 Female 779.14 621.44 669.80 1.20 0.55 NonHVR Male 733.20 579.06 564.14 2.80 0.25 Female 546.02 528.14 427.30 2.80 0.52 R-1 = pre-singing recording; R-2 = recording after 10 songs; R-4 = recording after subject reported vocal fatigue HVR group with Hydration (100 ml water) and Vocal Rest (one-minute) given between every song; NonHVR group non- Hydration and non-vocal Rest group *Significant at 0.05 level Hydration and vocal rests are frequently used as intervention strategies to restore vocal function or to improve voice quality. For example, Chan 28 used hydration and suggested frequent pauses during speaking to help improving the voice of kindergarten teachers. Similarly, Verdolini-Marston et al 29 reported improvement in voice and endoscopic examination following five days of hydration therapy in six females patients with laryngeal nodules or polyps. Solomon and DiMattia 13 also showed that phonation threshold pressure returned to baseline level following 15 minutes of voice rest. Together with the findings from the present study, hydration and vocal rest could certainly be recommended for karaoke singers as preventive measures of vocal fatigue.

226 EDWIN YIU AND RAINY CHAN However, there were several limitations with the methodology that caution should be exercised in attempting to generalize the results. The first limitation related to the subject selection. The number of subjects studied was small. There were only five subjects in each gender group and this small number did not allow us to use the more powerful parametric statistics to determine if there were differences in vocal fatigue between genders. In a previous study by Smith, Kirchner, Taylor, Hoffman and Lemke, 30 they found female teachers to report significantly more vocal discomfort symptoms, poorer vocal quality, and a higher rate of tired voice than male teachers. Although the findings from our study showed a change in jitter in the male subjects and a change in maximum pitch in the female subjects, these were not sufficient to enable us to determine whether there was a difference between the male and female voices in vocal fatigue. It should also be noted that the subjects recruited for the present study were restricted to young adults between 20 to 23 years of age. Their susceptibility to vocal fatigue might have been different from those of different ages resulting from the possible differences in anatomy and physiology of larynx as a result of aging. Therefore, future studies with larger groups of subjects, and with different ages and genders, will be needed to answer research questions on vocal fatigue related to age and gender. The second limitation related to the experimental condition. The giving of water and vocal rests simultaneously to the HVR group did not allow us to determine the effect of these two factors independently. In future, the use of two groups of experimental subjects, one with hydration only and the other with vocal rests only, would help to show the effects of these two factors separately. A related issue was the effect of alcohol and tobacco on vocal fatigue. Normally, karaoke singing activities are associated with large quantity of alcohol and tobacco consumption. In the present study, only subjects who were nonsmokers and nondrinkers were recruited. This should not be taken as a good representation of the real-life situation. If the real-life effect of karaoke singing and other associated factors on vocal fatigue are to be investigated, future studies will need to examine the alcohol and tobacco variables. Furthermore, the R2 examination required a 45- minute break in the singing routine. Although the phonetogram itself is a demanding task, the break from the singing routine could have led to some vocal improvement by allowing the voice to perform a slightly less demanding task than the singing task. This could be considered as some kind of rest. Despite these limitations, the findings from the study are useful for clinical purposes. As karaoke singing is a very popular entertainment among young people who usually do not have proper singing training, information on vocal hygiene would be particularly important to them in order to prevent the development of voice problems following prolonged singing. In this study, the experimental environment was already quieter than an average room in karaoke lounges. Furthermore, there was no smoking, no alcohol drinking and no snacks, which could have further irritated the voice. Therefore, the karaoke singers in the present study could have probably sung longer than average karaoke singers in a real karaoke lounge. Thus, the total amount of singing before feeling vocal fatigue revealed in the present study (ie, 85 minutes) should be taken as a guideline for karaoke singers who would like to avoid damaging their voice. In addition, singers should receive frequent hydration and vocal rests during singing to reduce the negative effect of prolonged voice use and delay the development of vocal fatigue. Acknowledgement: This study was supported in part by a grant from the Committee on Research and Conference Grant, University of Hong Kong (#10202073). We are grateful to Karen Chan and Chi Yan Ng, who served as judges for the perceptual analysis task. REFERENCES 1. Stone RE, Sharf DJ. Vocal changes associated with the use of atypical pitch and intensity levels. Folia Phoniatr. 1973;25:91 103. 2. Akerlund L, Gramming P, Sundberg J. Phonetogram and averages of sound pressure levels and fundamental frequencies of speech: comparison between female singers and nonsingers. J Voice. 1992;6(1):55 63. 3. Gelfer MG, Andrews ML, Schmidt CP. Effects of prolonged loud reading on selected measures of vocal function in trained and untrained singers. J Voice. 1991;5:158 167.

VOCAL FATIGUE IN AMATEUR SINGERS 227 4. Sander EK, Ripich DE. Vocal fatigue. Ann Otol Rhinol Laryngol. 1983;9:141 145. 5. Stemple J, Stanley J, Lee L. Objective measures of voice production in normal subjects following prolonged voice use. J Voice. 1995;9:127 133. 6. Eustace CS, Stemple JC, Lee L. Objective measures of voice production in patients complaining of laryngeal fatigue. J Voice. 1996;10:146 154. 7. Gotaas C, Starr CD. Vocal fatigue among teachers. Folia Phoniatr Logoped. 1993;45:120 l29. 8. Kitch JA, Oates J. The perceptual features of vocal fatigue as self-reported by a group of actors and singers. J Voice. 1994;8:207 214. 9. Long J, Williford HN, Olson MS, Wolfe V. Voice problems and risk factors among aerobics instructors. J Voice. 1998;12:197 207. 10. Koufman JA, Blalock PD. Vocal fatigue and dysphonia in the professional voice users: Bogart Bacall syndrome. Laryngoscope. 1988;98:493 498. 11. Novak A, Dlouha O, Capkova B, Vohradnik M. Voice fatigue after theater performance in actors. Folia Phoniatr. 1991;43:74 78. 12. Kostyk BE, Rochet AP. Laryngeal airway resistance in teachers with vocal fatigue: a preliminary study. J Voice. 1998;12:287 299. 13. Solomon NP, DiMattia MS. Effects of a vocally fatiguing task and systemic hydration on phonation threshold pressure. J Voice. 2000;14:341 362. 14. Vilkman E, Lauri ER, Alku P, Sala E, Sihvo M. Effects of prolonged oral reading on F 0, SPL, subglottal pressure and amplitude characteristics of glottal flow waveforms. J Voice. 1999;13:303 312. 15. Scherer RC, Titze IR, Raphael BN, et al. Vocal fatigue in a trained and an untrained voice user. In: Baer T, Sasaki C, Harris K, eds. Laryngeal Function in phonation and respiration. San Diego, CA: Singular Publishing; 1991: 533 555. 16. Mann EA, McClean MD, Gurevich-Uvena J, et al. The effects of excessive vocalization on acoustic and videostroboscopic measures of vocal fold condition. J Voice. 1999;13:294 302. 17. McAllister A, Sederholm E, Sundberg J, Gramming P. Relations between voice range profiles and physiological and perceptual voice characteristics in ten-year-old children. J Voice. 1994;8:230 239. 18. Van Mersbergen MR, Verdolini K, Titze IR. Time-of-day effects on voice range profile performance in young, vocally untrained adult females. J Voice. 1999;13:518 528. 19. Yiu E, Worrall LE, Longland J, Mitchell C. Analysing vocal quality of connected speech using Kay s Computerized Speech Lab: a preliminary finding. Clin Linguist Phonet. 2000;14:295 305. 20. De Bodt MS, Wuyts FL, Van de Heyning PH, Croux C. Testretest study of the GRBAS scale: influence of experience and professional background on perceptual rating of voice quality. J Voice. 1997;11(1):74 80. 21. Dejonckere PH, Obbens C, De Moor GM, Wieneke GH. Perceptual evaluation of dysphonia: reliability and relevance. Folia Phoniatr. 1993;45:76 83. 22. Hirano M. Clinical Examination of Voice. Springer Verlag: Vienna, Austria; 1981. 23. Laver J. The Phonetic Description of Voice Quality. Cambridge, U.K.: Cambridge University Press; 1980. 24. Chan KM-K, Yiu EML. The effect of anchors and training on the reliability of perceptual voice evaluation. J Speech Lang Hear Res. 2002;45:111 126. 25. Mehta CR, Patel NR. SPSS Exact Tests 7.0. for Windows. Chicago, IL: SPSS Inc; 1996. 26. Titze IR. Workshop on acoustic voice analysis: summary statement. Iowa City, IA: National Center for Voice and Speech; 1995. 27. Yiu EM-L. Limitations of perturbation measures in clinical acoustic voice analysis. Asia Pacif J Speech Lang Hear. 1999;4:157 168. 28. Chan RWK. Does the voice improve with vocal hygiene education? A study of some instrumental voice measures in a group of kindergarten teachers. J Voice. 1994;8:279 291. 29. Verdolini-Marston K, Sandage M, Titze R. Effect of hydration treatments on laryngeal nodules and polyps and related voice measures. J Voice. 1994;8(1):30 47. 30. Smith E, Kirchner HL, Taylor M, Hoffman H, Lemeke J. Voice problems among teachers: difference by gender and teaching characteristics. J Voice. 1998;12:328 334.