When listeners hear a voice, they hear a person and thus rapidly form a complex impression of the person behind that voice. We characterise how multivariate impressions from voices emerge over time across different levels of abstraction using electroencephalography (EEG) and representational similarity analysis (RSA). We find that for 8 perceived physical (gender, age, health), trait (attractiveness, dominance, trustworthiness), and social characteristics (educatedness, professionalism), representations emerge early (~80ms after stimulus onset), with voice acoustics contributing to those representations between ~100ms-400ms. While impressions of person characteristics are highly correlated, we can find evidence for highly abstracted, independent representations of individual person characteristics. For these abstracted representations, a staggered time course emerges, where representations of physical characteristics (age, gender) arise early (~120ms), while representations of some trait and social characteristics emerge later (~260ms onwards). The findings align with recent theoretical models and shed light on mechanisms underpinning person perception from voices.