Measure of speech replay accuracy during retrieval. A, During successful retrieval, participants recalled the auditory features associated with the speaker’s face, theoretically leading to mentally replaying “verbatim” the speech memorized during movie encoding. B, In this case, the brain activity in the auditory cortex (MEG; black line) reinstates the theta oscillatory patterns conveyed by the auditory speech envelope of the movie (speech envelope; orange line). C, The accuracy of speech reinstatement is measured with the level of phase similarity between the theta oscillation of the stimulus auditory envelope, and the neural theta oscillation produced replay of speech memory. To do that, the phase of the dominant theta frequency was extracted from the auditory envelope of the stimulus (orange line) and the MEG signal at the auditory sensors (black line). D, The phase similarity was estimated between the first sliding window (sw) of 1 s containing the theta phase of the auditory envelope centered on its first time-point (sw_t1) and every time-point of the MEG signal (top). The phase similarity was represented with a single value ranging between 0 and 1 at each time-point of the MEG signal (bottom). E, This operation was repeated by shifting the sliding window to the next time-point (sw_t2) of the auditory envelope signal and so on until its last time-point (sw_tn), in order to compute the phase similarity between the two signal dimensions over time (left). Theta phase similarity was then averaged across windows, trials, and sensors of interest for statistical analyses (right). Adapted from Michelmann et al. (2016).