This paper presents a comparative analysis of the results in the prediction of the Speech Transmission Index (STI), with a focus on estimation error and prediction accuracy. STI prediction was performed using several regression equations applied to Room Impulse Responses (RIRs) recorded in the lecture hall of the Academy of Applied Technical and Preschool Studies, Niš Department, Serbia. Based on the RIRs, the values of reverberation time (RT) and STI, at the central frequencies of the octave band, fc = {250, 500, 1000, 2000, 4000} Hz, were calculated. Using the calculated RTs values, the STI, through linear regression, was estimated and, in this paper, denoted as VSTI. Additionally, STI was estimated using regression equations (TSTI, ElinSTI, ElnSTI), which were described in scientific literature. To assess the precision of the estimation of the VSTIs using the regression equations, a comparative analysis, which were based on statistical parameters (µ, SD, σ2), was conducted. The experimental results are presented using tables and graphs. A detailed comparative analysis of the estimation error in STI prediction revealed that the lowest prediction error was obtained using linear regression VSTI and the ElinSTI equation.