When judging the time-to-collision (TTC) of visually presented accelerating vehicles, pedestrians do not adequately consider the acceleration and commit systematic estimation errors (first-order pattern, resulting in TTC overestimation), which represents a potential risk. These estimation errors are largely removed when the sound of conventional (but not electric) vehicles is presented. We investigated whether a visual acceleration signal (Exp 1.) or trial-by-trial feedback about the estimation accuracy (Exp. 2) can improve visual-only TTC estimation for accelerating vehicles and thus can reconcile pedestrian safety and noise reduction. In a VR traffic simulation showing a pedestrian's perspective at the curb, participants estimated the TTC for constant-velocity and accelerated vehicle approaches. In half of the blocks in Exp. 1 (n = 26), a light band on the windshield illuminated whenever the vehicle accelerated but remained deactivated at constant speeds. In the other blocks, the light band never illuminated. Without acceleration signal, we observed the expected erroneous, first-order pattern in TTC estimation for accelerated approaches. With acceleration signal, the first-order pattern was reduced. Exp. 2 (n = 20) included 3 blocks, of which only the second provided trial-by-trial feedback about the TTC estimation accuracy. Although participants adjusted their estimations during/after the feedback, in each block, the estimations for accelerated approaches showed a consistent first-order pattern. This suggests that the feedback did not help participants in considering the acceleration. Overall, a binary visual acceleration signal rather than training with feedback could be a potential countermeasure against pedestrians' erroneous TTC estimation for silent accelerating vehicles.