Recent work has suggested that bilingual listeners use the visual identity of the talker to form expectations about the language the talker will use, which then facilitates lexical processing. In the current study, we extend this work to see if there are analogous effects of talker identity on dialect processing, and whether the impact of talker identity depends on the regional background of the listener. Six actresses recorded stimuli in two dialectal guises, performing Southern US accents and standardized, regionally nonspecific US accents. Participants were introduced to the actresses via video as having one particular dialect type (familiarization), and then later did an audio-visual lexical decision task (test) where some trials would be dialectally congruent and some trials would be dialectally incongruent with their earlier experience of that talker. US English listeners from both Southern and non-Southern dialect regions participated. Listeners who self-reported having (Southern) accents were impacted by talker dialect congruence, performing best with a given dialect when it matched their experience of that talker. However, other listeners were not impacted by congruency, performing better with standardized tokens regardless. This mirrors findings in bilingualism research that early bilinguals are more sensitive to talker language pairing than monolinguals or late bilinguals. We ran three additional conditions without video and/or without a familiarization stage to confirm the importance of each component to observing the effect. Generally, without familiarization, Southern US listeners performed worse with Southern vs. standardized tokens, suggesting that without strong contextual cues indicating otherwise, these listeners may expect standardized tokens in experimental settings. There is some evidence that all listeners were somewhat sensitive to talker identity even from voice alone.