This study explored perceptual sensitivity to spectral and temporal acoustic features in music recognition among BaYaka hunter-gatherer children. Using an interactive, non-competitive game (Move to Match), we tested song recognition under various forms of audio degradation to identify which musical dimensions are crucial for recognizing familiar songs. BaYaka children demonstrated robust recognition of songs across unaltered, spectrally degraded, and temporally degraded versions, highlighting sensitivity to multiple acoustic dimensions. We suggest that songs may be recognized despite degradations because of spectral contour processing, rather than pitch and rhythm cues alone. Moreover, recognition performance decreased for duration-degraded songs (“plinks”), as the brevity constrains the cues available to listeners. However, song identification of (spectrally degraded) plinks far exceeded chance levels and outperformed Western adults performing similar tasks. This suggests that onset-transients and spectral shape, contributing to timbre perception, might serve as reliable markers for identifying musical sounds. We discuss the possibility that music cognition evolved for reasons beyond social bonding to support identification of personal contributions within musical exchanges through timbre, which may hold importance in cooperative contexts. As one of the first experimental studies of music perception in hunter-gatherers, this work demonstrates the potential of culturally adapted methods for future cross-cultural research.