The article proposes a modification of the methodology for assessing the quality of pronunciation of a patient’s syllables in the process of speech rehabilitation. A modification of the technique consists in applying an approach that allows the use of recordings of healthy speakers to assess the quality of pronunciation of various phonemes. To assess the quality of pronunciation, we use the distance between two files with recordings in wav format; the smaller the distance between two files, the more similar they are considered. Distance calculation methods discussed included DTW, EDR, ERP, LCSS, MSM, and Euclidean distance. The accuracy of the method was tested using values ​​obtained using the instantaneous energy envelope and the Gilbert envelope, in the first case ERP was the most accurate method, in the second EDR. It was revealed that the LCSS and ERP methods perform calculations much longer than other methods and do not have high accuracy, as a result, they are not included in the list of the best calculation methods. The hypothesis that the two means were equal was also tested using the Mann-Whitney method to confirm that the distance between the healthy speaker recording and the preoperative patient recording differs from the distance between the healthy speaker recording and the recording of the patient after surgery. As a result, it turned out that the most accurate and fastest method for calculating the distance between two records is EDR. Analysis of the results obtained showed that this approach is applicable to solving the problem of speech analysis, but requires significant improvements and subsequent research due to the low accuracy of the put forward theories.