Comparison of Praat and Computerized Speech Lab for formant analysis of five Japanese vowels in maxillectomy patients

Abstract

IntroductionSpeech impairment is a common complication after surgical resection of maxillary tumors. Maxillofacial prosthodontists play a critical role in restoring this function so that affected patients can enjoy better lives. For that purpose, several acoustic software packages have been used for speech evaluation, among which Computerized Speech Lab (CSL) and Praat are widely used in clinical and research contexts. Although CSL is a commercial product, Praat is freely available on the internet and can be used by patients and clinicians to practice several therapy goals. Therefore, this study aimed to determine if both software produced comparable results for the first two formant frequencies (F1 and F2) and their respective formant ranges obtained from the same voice samples from Japanese participants with maxillectomy defects.MethodsCSL was used as a reference to evaluate the accuracy of Praat with both the default and newly proposed adjusted settings. Thirty-seven participants were enrolled in this study for formant analysis of the five Japanese vowels (a/i/u/e/o) using CSL and Praat. Spearman’s rank correlation coefficient was used to judge the correlation between the analysis results of both programs regarding F1 and F2 and their respective formant ranges.ResultsAs the findings pointed out, highly positive correlations between both software were found for all acoustic features and all Praat settings.DiscussionThe strong correlations between the results of both CSL and Praat suggest that both programs may have similar decision strategies for atypical speech and for both sexes. This study highlights that the default settings in Praat can be used for formant analysis in maxillectomy patients with predictable accuracy. The proposed adjusted settings in Praat can yield more accurate results for formant analysis of atypical speech in maxillectomy cases when the examiner cannot precisely locate the formant frequencies using the default settings or confirm analysis results obtained using CSL.