On listening to this woman’s two pronunciations of “street” and inspecting the waveforms and spectrograms, it seems that the two vocoids are plainly different. The one between the [s] and the [t] is much longer and has a reasonably clear quality, close to barred i, i.e. backer than cap I, as one of the previous posters noted. The one between the [t] and the tap [r] is much briefer, indistinct in quality and indeed hard even to detect by ear. These differences suggest that the first is actually intended, and not a product of minimal overlap between the [s] and [t], while the second is a mere byproduct of the lack of overlap between the [t] and the tap.