Sound frequency analysis and the site of snoring in natural and sedation-induced sleep

Abstract

Frequency analysis of snoring sounds has been shown to correlate with the site of upper airway narrowing. The aim of this study was to compare the sound frequency spectra of snores generated during induced snoring at sleep nasendoscopy with those during natural sleep. The sound of 11 subjects snoring overnight and during sleep nasendoscopy was digitally recorded. The observed site of vibration was noted for each snore sample recorded at sleep nasendoscopy. A Fast Fourrier Transformation was used to produce frequency spectra. At sleep nasendoscopy observed palatal snores (76) had a median peak frequency at 136 Hz, tongue based snores (10) at 1243 Hz and simultaneous palate and tongue (six) at 190 Hz. The median power ratios (defined as the area under the frequency spectra below 750 Hz/area above 750 Hz) were 33 for palatal snores, 0.2 for tongue base snores and 5 for simultaneous palate and tongue. Natural snores (275) had a dominant median peak at 132 Hz with a ratio of 88. A comparison of overnight snores (200) of those subjects (eight) with observed palatal snores (76) was made. Neither the peak frequency nor power ratio of natural and induced snores were significantly different (P = 0.85 and P = 0.13, respectively). The power ratio in four of the subjects individually was significantly different (P < 0.05). These results suggest that sedation-induced snores contain a high frequency component consistent with an element of tongue base snoring, not evident in natural snores. Thus sleep nasendoscopy may not accurately reflect snoring during natural sleep. Sound frequency analysis during natural sleep may therefore be a better predictive investigation for the success of palatal surgery.