the probe that is reflected from the interfaces of the sample and the light partially reflected from the strain pulse. At normal incidence, the period can be written as T ¼ k=2nv ,w herek is the probe wavelength, n is the optical index, and v is the sound velocity. By measuring the oscillation period over a wavelength range, the authors of Ref. 1 deduced the longitudinal sound velocity at various acoustic frequencies. Surprisingly, they found that the sound velocity is not constant but varies by 5% in the investigated phonon wave vector range. The intent of this Comment is to warn readers that the conclusion is probably wrong due to the setup itself. We obtained new data which are very close to what is expected from a substrate, and no deviation of the sound velocity is observed. We performed experiments on a 10 � 10 � 0.5 mm 3 bulk commercial (100) STO substrate 3 As in Ref. 1, the substrate was covered by a thin Al film in order to convert light pulses into acoustic pulses. We used a conventional pump and probe setup associated with a femtosecond laser tunable between 690 and 1040nm at a repetition rate of 80MHz with 120 fs optical pulse width. The probe is frequency doubled to reach the blue-UV range. Our data are obtained by tuning the laser wavelength by wavelength. The signal to noise ratio is better than using a spectroscopic setup which permits to study the response of the substrate over a wider wavelength range. In Fig. 1, we plot our experimental data, the data from Ref. 1 and the theoretical data obtained from the Brillouin formula, sound velocity, and optical index are extracted from lit