Using a closely sampled 1000-km hydrographic section in the eastern North Pacific, the sound-speed finestructure is separated into two component fields: (i) isopyncal tilt dominated by internal waves (the traditional view) and (ii) "spicy" (cold-fresh to hot-salty) millifronts associated with upper ocean stirring. Numerical transmission experiments show significant scatter within the mixed layer from the spicy fronts. This scattered energy arrives near the start of the SOFAR sequence, and is superimposed on a triplication of the channel dispersion at the transition from reflected to upper ocean refracted energy. (This SOFAR overture is totally different from the finale which has been prominent for over 50 years.) The critical dependence of the overture on mixed layer processes suggests a scheme for acoustically monitoring the upper oceans at surface-conjugate depths (3 to 5 km), offering some advantages over in situ monitoring. (C) 2004 Acoustical Society of America.