Several studies show that pre-sleep learning determines changes in subsequent sleep, including improvements of sleep quality. Our aims were to confirm this finding using a more ecological task (learning a theatrical monologue) and to investigate whether the effect is modulated by expertise. Using a mixed design, we compared polysomnographic recordings of baseline sleep (BL, 9-h TIB) to those of post-training sleep (TR, with the same TIB but preceded by the training session), in one group of actors (N = 11) and one of non-actors (N = 11). In both groups, TR appears reorganized and re-compacted by the learning session, as shown, among others, by a significant decrease of WASO%, awakenings, arousals, and state transitions and by a trend towards an increased number of complete cycles and total cycle time. Concerning memory performance, the number of synonyms produced was significantly higher in the morning relative to immediate recall. No between-groups differences emerged either for sleep or memory variables. Our data confirm pre-sleep learning’s beneficial effect on sleep quality in an ecological context. While expertise appears not to influence memory-related sleep mechanisms, results on morning recall support the recent view that sleep’s role in memory processes consists in trace “transformation” for adaptive purposes, rather than rote consolidation.