Measurements of jitter and shimmer are frequently employed to quantify laryngeal control and stability during voice production in patients(1-3). Accurate estimates of jitter and shimmer may be obtained from acoustic recordings of sustained vowels produced by patients using computer software implementing algorithms for their extraction (e.g. Praat(4) or MDVP(5)). The jitter and shimmer algorithms do not, in themselves, exclude non-stable productions and are influenced by the inclusion of silence in the analysed intervals. As a consequence, reliable measurements of jitter and shimmer are made by manually opening each sound file and selecting an interval for the computations.
The current paper proposes an alternative approach to afford a more efficient estimation of jitter and shimmer across a large set of sustained vowel recordings. Using information readily available in the acoustic signal and a combination of algorithms already available within the Praat program, a reliable method for automatic processing of only the sustained vowel in each recording of a large corpus is outlined. The method further affords the acquisition of multiple, repeatable, measurements of jitter and shimmer for sub-intervals of the vowel’s duration (applying more than one algorithm), which additionally provides information concerning the reliability of the jitter or shimmer estimates for a specific vowel production. Comparisons with manually obtained measurements are made for the purpose of validation of the segmentation method.