Editorial board editor-in-chief

Fig. 1.
Short time Fourier transform
Fig. 2.
Normalized waveform of three snoring events over
a period of 10s

---

39
T
elema
tics
Snoring as a sign of abnormality
Fig. 3.
The averaged spectrum shape of two snoring events 
(left: second snoring event, right: third snoring event) for a single patient
Fig. 4.
Time variation of the frequency spectrum of two snoring events
(left: second snoring event, right: third snoring event) of a single patient
Frequency, bandwidth and amplitude levels characterize 
formants. The range of these parameters depends on the shape 
of the resonant cavities. According to Perez–Padilla et al[5], three 
types of snoring can be distinguished, namely oral, nasal and 
oronasal, which present different spectra. Acoustic analysis of 
vowels revealed decreased values of F3 and F4 formant levels 
and energy concentration in the lower range of frequencies, sug-
gesting the presence of acoustic characteristics of nasal snoring. 
This is the most common resonance disorder.
Thanks to a complete acoustic analysis we are able to extract 
two- or three-dimensional acoustic images which represent time 
variability in each frequency band. This kind of presentation shows 
three types of information: in the time and frequency domain and 
amplitude sound level. In igure 5 normal and pathological signals 
are compared. It is noteworthy that the pathological signal has 
more transient components, especially at lower frequencies.

---

T
elema
tics
40
Snoring as a sign of abnormality
the fundamental frequency, moments M0-M2 and formants. 
The pathological importance of snoring has been related to its 
intensity (dB), maximal and mean intensity, number of breaths 
per minute of sleep, snoring frequency and formants structure. As 
it is presented in table 2 below, the mean energy is signiicantly 
higher for habitual snorers when compared to occasional snorers.
The signal of a snorer presents weak formants, while the 
normal signal has more periodicity in low frequencies, and in-
troduces stronger formants. Calculated parameters which form 
a vector of features of the abnormal sound were quantitatively 
compared with normal sound, which allowed drawing conclusions 
from the available data. The most important parameters were