Acoustic travel-time tomography (ATOM) is a widely used experimental technique for remote sensing of indoor air temperature. The procedure of ATOM corresponds to collecting the time-of-flight of early reflections in room impulse response, which is then used as the input data to reconstruct temperature within the tomographic area. Usually, the coordinates of the used transceivers are known; however, practical constraints often prevent the accurate access of these coordinates, which can have an impact on the measurement accuracy. In the present work, we address the issue of predicting accurate coordinates of transceivers. Unlike earlier published papers by Dokhanchi et al. [Measurement 164, 107934 (2020)], which are based on adjusting the distance between transceivers, we propose a robust and accurate calibration method. Results show the feasibility of the proposed method, where 109 updated coordinates out of the total 201 successfully recover a highly accurate temperature within a plexiglass box (1.33 m × 1.0 m × 1.27 m). Compared to the method based on the adjustment of the distance between the transceivers, the present calibration method reduces the uniform temperature discrepancy from 0.22 °C to 0.05 °C. Moreover, we show that this calibration method remains applicable for indoor graded temperature.