Professional Documents
Culture Documents
Acoustical Comparison Between Samples of Good and Poor Vibrato in Singers - 2003 - Journal of Voice
Acoustical Comparison Between Samples of Good and Poor Vibrato in Singers - 2003 - Journal of Voice
179
180 JOSE A. DIAZ AND HOWARD B. ROTHMAN
FIGURE 1. Spectrogram of vibrato sample. which was developed by one of the authors, was
selected for sample analysis because it provided
SAMPLE SELECTION the time varying rate and extent in Hertz of the
frequency vibrato wave. This software generated
For this research, eight professional singers (5 sample spectrograms (see Figure 1), from which the
male, 3 female) with years of experience were vibrato samples were extracted. Figure 2 shows
chosen from a database of vibrato samples that is kept the frequency vibrato wave extracted from the sixth
at the Department of Communication Sciences and harmonic in Figure 1.
Disorders at the University of Florida. Four different The software used the LPC method to generate the
individuals with singing experience or with experi-
time varying rate and extent waves of the frequency
ence in studying the singing voice listened to all
vibrato, which represent how the rate and extent of
samples from the eight singers in order to classify
the frequency vibrato wave vary along the time axis.
the samples produced by each singer as samples of
good or poor vibrato. The judges were instructed to These two waves were created by taking a small
classify the samples depending on their quality, as segment of the vibrato signal (170 ms.), applying
good (pleasing to the ear) or poor (disagreeable to the LPC method, and obtaining the rate and extent
the ear). in Hertz of the segment from the LPC parameters
The results provided by each judge were com- obtained. Then, the software selected a new segment
pared, and the samples that were judged as good of the same length (170 ms.), shifted 12 ms. to the
or poor by the four judges were chosen for analysis; right and analyzed it. This process was repeated
the other samples were rejected because the until the entire signal was analyzed.
judges did not unanimously agree on their quality. The length of the vibrato segment was chosen to
From the chosen samples, five samples of good vi- be equal to 170 ms. because a longer segment would
brato and five of poor vibrato were randomly se- show the average rate and extent of the segment,
lected for each singer, which resulted in a total of and a shorter segment would not contain enough
80 samples (8 singers times 10 samples/singer). All data for a reliable result. Then, all rate and extent
samples selected fulfilled the following criteria: values were plotted on separate windows (see Fig-
1. Length greater than 1.5 s. ures 3 and 4). These two waves represent the time
2. No change in pitch. varying rate and extent of the frequency vibrato
3. No change in vowel. wave seen in Figure 2. Notice that the time scale
4. Absence of accompanying musical instru- in Figures 3 and 4 is shorter than that of Figures 1
ments or noise. and 2. This is because the LPC method requires a
segment of the vibrato wave to calculate one value
SAMPLE ANALYSIS of the rate and extent. It is important to note that
The MMSV1 (Mathematical Model of Singers’ the time varying extent wave represents the extent
Vibrato, The Mathworks, Natick, MA) software, of the wave in Figure 2, and it does not represent
for α = 0.05, df1 = 1, and df2 = 8). Columns 4 and the variables under study were higher for the poor
5 show the F and P values respectively. No arc- samples in four singers of eight. No significant re-
sine transform was performed on the data. Column 6 sults were found for singers numbered 1, 4, 6, and
indicates whether the obtained results were significant. 8. These results support the hypothesis that the best
Significant results were found in four of the eight samples were the most symmetrical ones and serve
singers, which indicates that the mean values of to validate the listeners’ classification of the samples.
Analysis of the vibrato rate for the of the variability of the time varying extent yielded
individual singers significant results in four singers of eight, whereas
We can see from Tables 1 through 8 that the the total deviation of the time varying extent yielded
variability in rate for the poor samples was always significant results in three cases of eight.
higher than that of the good samples. The total devia-
tion of the time varying rate was higher in seven
of the eight singers and was only smaller for singer Comparison between good and poor samples
7. The higher total deviation of the time varying for the entire group
rate for the good samples in singer 7 indicates large Table 9 provides a summary of the results after
oscillations of the time varying rate, which were not comparing the good and poor samples of the entire
perceived as poor vibrato. The statistical analysis group. The first column shows the parameters
of the variability of the time varying rate did not under analysis. Columns 2 and 3 show the average
show significant results, nor did the total deviation results for good and poor samples. Single factor-
of the time varying rate. ANOVAs with eight samples per group (8 singers)
and two groups (good and poor samples) were ap-
Analysis of the vibrato extent for the plied to the four parameters under analysis (critical
individual singers value of F = 4.60 for α = 0.05, df1 = 1, and df2 =
We can see from Tables 1 through 8 that the 14). Columns 4 and 5 show the F and p values
variability and the total deviation of the time varying respectively. No arc-sine transform was performed
extent for the poor samples were always higher than on the data. Column 6 indicates whether the ob-
that of the good samples. The statistical analysis tained results were significant or not.
Analysis of the vibrato rate for the It could also be seen that the oscillations in the
entire group rate and extent of frequency vibrato wave were
Table 9 indicates that the mean values of the found mainly in the time varying extent wave, rather
four parameters were always smaller for the good than in the time varying rate wave, which shows that
samples than for those of the poor samples. In Table it was more difficult for singers to control the extent of
9, the variability and the total deviation of the time the vibrato pulse than the rate at which the vibrato
varying rate do not show significant results. How- pulse occurred. Taken together, these results show
ever, it is important to note that the poor samples a direct relationship between the periodicity of the
used in this study showed higher oscillations in rate vibrato wave and its perceived quality.
than that of the good samples, although no significant
results were found. REFERENCES
1. Diaz JA. Frequency characterization of singers’ vibrato.
Analysis of the vibrato extent for the Gainesville, FL: University of Florida; 1995. Master’s thesis.
entire group 2. Sundberg J. The Science of the Singing Voice. Dekalb, IL:
In Table 9, the variability and the total deviation Northern Illinois University Press; 1987.
of the time varying extent show significant results. 3. DeJonckere PH, Minoru H, Sundberg J. Vibrato. San Diego,
California: Singular Publishing Group; 1995.
Therefore, it appeared that the oscillations observed 4. Prame E. Measurement of the vibrato rate of ten singers.
in the rate and extent of the frequency vibrato wave J Acoust Soc Am. 1994;96:1979–1984.
were mainly due to variations in extent. 5. Horii Y. Frequency modulation characteristics of sustained
/a/ sung in vocal vibrato. J Speech Hear Res. 1989;32:1–8.
6. Maher R, Beauchamp J. An investigation of vocal vibrato
for synthesis. Appl Acoustics. 1990;30:19–245.
CONCLUSIONS 7. Sundberg J. Acoustic and psychoacoustic aspects of vocal
vibrato. KTH Speech Transmission Lab Quart Progr Status
The results of this research support the hypothesis Report. 1994;2–3:45–48.
that the most symmetrical samples were judged 8. Titze IR. Synthesis of sung vowels using a time-domain
as good samples of vibrato. Also, it appeared that approach. Transcripts of the Eleventh Symposium: Care of
the largest number of significant results was found the Professional Voice. New York: The Voice Foundation;
in the variability of the time varying extent (4 cases 1983:90–98.
9. Mellody M, Wakefield G. The time-frequency character-
of 8), which suggests that the variability of the time
istics of violin vibrato: modal distribution analysis and
varying extent provided a measure of vibrato quality synthesis. J Acoust Soc Am. 2000;107:598–611.
that was more accurate and meaningful than the 10. Ott RL. An Introduction to Statistical Methods and Data
other three parameters under analysis. Analysis. Belmont, CA: Duxbury Press; 1993.