We study the mixing of active scalars by homogeneous isotropic incompressible stochastic velocity fields. We consider both Navier-Stokes generated turbulent fields as well as artificially generated homogeneous isotropic stochastic fields. We use Fourier pseudospectral direct numerical simulations to study the mixing dynamics of two non-reacting species of different density ratios. We use the Atwood number to create a denser mixture and a lighter mixture. We show that in the absence of stirring, a denser mixture homogenizes faster than the lighter mixture. The direction of the density gradient causes the interface across which the molecular diffusion occurs to expand outward for the denser mixture and inward for the lighter mixture. The stirring process, which enhances the diffusion process, increases the rate of homogenization in both mixing methods under study. We define a new mixing metric for studying the mixing evolution of active scalars, which indicates that a denser inhomogeneity in a lighter mixture spreads faster but homogenizes slower. For low Mach number turbulence, there is a negligible coupling between the density gradients and the velocity field responsible for stirring. The post-stirring behavior of active scalars is found to be similar to passive scalars, where the scalar energy spectra decay exponentially and exhibit self-similarity. The turbulence fields generated by solving the Navier-Stokes equation homogenize both the mixtures faster than the synthetic cases. We show that matching the kinetic energy spectra and inertial subrange scaling of a synthetically generated stochastic field with that of a Navier-Stokes generated field is not enough to study mixing dynamics.