Scientific Papers

Variation of the acoustic parameters: f0, jitter, shimmer and alpha ratio in relation with different background noise levels

Speech is one of the pillars of human communication, so that any alteration in the process of speech production may have a negative effect.1 Many studies over the years have focussed on how to differentiate a pathological voice2 from a euphonic one.3

Voice evaluation methods have centred on two lines of analysis. The first line includes the use of scales such as the GRBAS, corresponding to the perceptual analysis of a voice and the creation of an overview of the vocal quality of an individual.4 One disadvantage of this type of evaluation is that it is subjective, given that it is determined by an examiner and the degree of their expertise.5 On the other hand there is the objective analysis obtained by processing a voice using acoustic analysis programmes such as Praat. This analysis is one of the most widely used to identify voice disorders,6 and it allows us to obtain information about certain acoustic parameters, such as: fundamental frequency (f0), Jitter, Shimmer, Harmonic to Noise Ratio (HNR) and the Alpha ratio, among others.

The f07 corresponds to the number of times the glottis opens and closes in the time period of 1 second. This parameter is linked to an individual’s tone of voice, and it varies depending on their sex and age,8 and it may also be affected by their mood9 or prosody.

The Jitter10 parameter evaluates perturbation of the fundamental frequency by comparing one cycle with another. This parameter is affected if the cycle of vibration of the vocal folds is not controlled. It is measured in [%], and higher values indicate less stable vibration cycles. Praat sets the normality threshold at 1.04%.11

The Shimmer parameter evaluates sound wave amplitude perturbation.12 This is measured in [%], and higher values indicate greater sound wave amplitude instability. Praat set the normality threshold here at 3.81%.11

The HNR parameter evaluates the relationship between the percentage of harmonics and the percentage of chaotic noise present in the audio analysed, corresponding to the voice,12, 13 so that if 99% corresponds to harmonics and only 1% corresponds to noise, the HNR value will therefore be 20 [dB]. On the other hand, if there are 50% harmonics and 50% noise, the HNR will be 0 [dB].11 This shows that lower HNR values indicates higher turbulent noise levels in the voice, and therefore a voice that is hoarse. In their 1982 study Yumoto, E. et al14 established that a HNR value below 7.4 [dB] is associated with a pathological voice.

The Alpha Ratio parameter is obtained from the long-term average spectrum (LTAS).15 It evaluates the relationship between low harmonics and those above 1000 [Hz], and it is associated with spectral slope,16 the delivery of information about the source as well as the filter17 and voice quality.18 Hypofunctional voices are associated with low Alpha ratio values, while hyperfunctional voices are associated with high Alpha Ratio values.19

To correctly perform an acoustic analysis it is vitally important to ensure that the process of making a voice recording is implemented correctly, given that any interference in this process may lead to erroneous acoustic parameter values that would cause the professional to commit diagnostic errors. It is clear that the background noise in the location where acoustic samples are taken must be as low as possible, and it is important to quantify this level, to set a maximum value for background noise that does not interfere with the values of the acoustic parameters. The 2021 study by Marsano-Cornejo, M. J. et al.20 found that significant differences arise in the HNR acoustic parameter when the background noise levels in a location exceeds 47.7 dB(A), so that they set a maximum level of 43.8 dB(A). This study does not offer information regarding what happens with the other acoustic parameters, i.e., whether significant differences will arise at the same background noise level in the location.

The purpose of this study is to continue the above-mentioned research and evaluate whether the f0, Jitter, Shimmer and Alpha Ratio acoustic parameters differ significantly at the same background noise level.

Source link