The THz initial spectrum of Fig. 3 refers to an infrared pump 10 fs long rather than 100 fs as erroneously reported at the second line of the second column of page 3. This correction is only made in order to give self-consistency to the paper. It doesn’t affect at all the validity of the introduced concepts and the range of applicability of the proposed diagnostic method. Furthermore, due to the high-frequency absorption of the quartz capillary, any pump length in the range 0–50 fs would produce the same numerical results as in Fig. 4. The choice of 10 fs is arbitrary and not much relevant for the purposes of the Letter.