From bars to business meeting, we are accustomed to engaging in conversation within noisy environments. Yet children and adolescents encounter more challenges than adults when it comes to understanding speech in noise. Speech intelligibility in noise is likely influenced by auditory segregation, an aspect of auditory scene analysis that remains immature at least until late childhood. Notably, musical abilities also seem to play a role in the development of speech perception in noise. The aim of this study was to investigate the respective contribution of stream segregation and perceptual musical abilities on the development of speech intelligibility in noise. In the present study, we recruited children (n = 80), adolescents (n=40) and adults (n=80) non-musicians with varying levels of perceptual musical abilities. Participants performed a stochastic-figure ground discrimination task which evaluates auditory segregation and a consonant identification in noise task. Using structural equation modelling, we observed a developmental improvement in auditory segregation and speech intelligibility in noise. Moreover, the relationship between musical abilities and speech perception in noisy environments appeared to be mediated by auditory segregation. Additionally, our results suggest a developmental improvement of the mechanisms of auditory scene analysis and speech intelligibility in noise, from childhood to adulthood. These observations are in line with recent views stating that music may have emerged as a cultural creation relying upon preexisting adaptations for auditory scene analysis.