Excited to read this final version - enjoyed the pre-print. Help me get your point here, is it not the case that acoustic features are spectral and temporal modulations? What does it mean in lay terms that song and speech differ in spectro-temporal peaks? One is faster? One has less spectral mod?

the general point is that the acoustic features differentiating the two types of vocalizations are structured, in the sense that the variability in sounds across speech and song is much more parsimoniously explained by spectral and temporal modulation than acoustic features which are made up of them