Spectral analysis of the voice in Down Syndrome

The voice quality of individuals with Down Syndrome (DS) is generally described as husky, monotonous and raucous. On the other hand, the voice of DS children is characterized by breathiness, roughness, and nasality and is typically low pitched. However, research on phonation and intonation in these participants is limited. The present study was designed to provide data from the spectral analysis of the human voice in DS people. A cross-sectional, observational design was applied. Thirty DS adults and 48 DS children were enrolled after clinical evaluation. Thirty men, 30 women and 46 children constituted the control group. The participants had to repeat a set of Italian words twice. The Real Time Pitch software manufactured by KayPENTAX recorded the voice. The following spectral descriptors were obtained for each word: Mean Frequency and standard deviation, Energy, Duration, Jitter and Shimmer. Test-retest performance was also checked. The voice of DS adults was characterized by a significantly higher Mean Frequency, particularly in males (p<0.0001), by a smaller variation (p=0.0044 in males and p=0.0046 in females) and by a significantly lower level of Energy (p=0.0037 in males and p=0.0025 females). Furthermore, limited to male adults, a shorter Duration (p=0.0156) and a smaller value of Shimmer (p=0.0014) was observed. The difference between DS children and age-matched controls was limited, reaching significance only for the Coefficient of Variation (CV) (p=0.031). The difference in Mean Frequency between adults and children was more evident in the control males than in all other groups. The lack of marked difference between voice characteristics of children with and without DS is outlined by findings. Pearson's correlation coefficients on repeated productions ranged from 0.23 (Jitter) to 0.86 (Mean Frequency) in children, and from 0.07 (Shimmer) to 0.86 (Mean Frequency) in adults. In the control group, all the coefficients ranged between 0.85 and 0.98. As expected, women had a higher Mean Frequency than men, but the CV was around 0.1 for both. By contrast, children had a significantly higher Mean Frequency and a lower CV. In conclusion, spectral analysis of the human voice is recommended in each laboratory of speech and language rehabilitation to exploit the accuracy of voice descriptors.