Klaske van Sluis

12 1.6. Multidimensional analysis of voice and speech is standardly performed with the GRBAS scale, including the five parame- ters Grade, Roughness, Breathiness, Asthenia, and Strain. Unfortunately, this scale is not sufficient for evaluating substitute voicing, as these largely irreg- ular voicing types often score the highest grade of dysphonia. For substitute voices, judgements of experienced SLP’s are considered as more proficient than judgements of naïve raters [24]. In the literature various variables are applied to evaluate substitute voices by perceptual ratings. Moerman and colleagues developed the IINFVo scale to assess substitute speech perceptually. The five IINFVo scale parameters are: overall impression (I), impression of intelligibility (I), unintended additive noise (N), fluency (F) and voicing (Vo) [25]. It appears that the variables can be reduced to two perceptual dimensions which are the most reliable: acceptability and intelligibility [25, 26]. Overall impression or acceptability of the voice refers to the degree to which speech is agreeable to the ear and has a pleasant quality. Intelligibility refers to the clarity and un- derstandability of words and sentences [25, 26]. 1.6.2 Acoustic analysis Acoustic analysis of the voice includes analysis of the signal of audio recordings of the voice and speech. Acoustic outcomes provide an objective measure of vocal functioning. The characteristics of the voice signal can be analyzed with the help of software; frequently used software is Praat [27]. Standard acous- tic measures of a sustained vowel /a/ include perturbation measures, providing information about the period-to-period and amplitude variation of the voice sig- nal, jitter and shimmer, respectively. A limitation of this perturbation analysis is that the systems often cannot analyze strongly aperiodic signals which are present in substitute voicing, therefore the validity and clinical utility of many of these acoustic measures has been strongly debated [28–30]. Thus, additional information can be provided by measuring the proportion of voiced frames and signal-to-noise ratio computations. Signal-to-noise ratio computations include harmonics-to-noise-ratio and cepstrum peak. Over the last decade automatic speech and voice analysis has become field of interest [30–32]. Automatic anal- ysis is promising in providing consistent ratings and analyzing trends within a single speaker. To perform an automatic analysis which combines several acoustic parameters in one score the Acoustic Voice Quality Index can be used [30]. The AVQI is a measurement of six acoustic domains in time, frequency and quefrency modeled in in a linear regression formula. 1.6.3 Patient-reported outcome measures Patient-reported outcome measures (PROM’s) are ratings performed by the patient; patients evaluate their own voice quality, understandability, and func- tioning with the voice. This evaluation is subjective by definition, but very important in clinical practice. It is the patient who has to function in everyday life with the changed voice; therefore these questionnaires often assess the diffi-