In this article the authors present the idea of automatic processing of singing recordings to create a database and for later analysis. The recordings contain sequences of sounds of the piano and singing separated by short segments of silence. The aim of the presented approach is the precise segmentation of the recordings and standardization of the resulting segments (termed further as samples) for later assessment of vocal quality. The taken approach can reduce the amount of time required for preparation of a database with sequences of singing for the analysis of singing voice quality. The samples obtained in the process must meet the prescribed requirements to enable carrying out the planned analyzes such as estimation of the fundamental frequency, vibrato estimation, Singing Power Ratio and formants estimation.