In general relativity, the continuity equation says $$ \partial_{\mu}\left(\rho_0c\dfrac{dx^{\mu}}{ds}\sqrt{-g}\right) = 0 $$
with $\rho_0$ being the proper density, as seen by an observer who is at rest with respect to the matter.
This is a general-relativistic version of $$ \partial_{\mu}\left(\dfrac{\rho_0}{\sqrt{1-\tfrac{v^2}{c^2}}}\dfrac{dx^{\mu}}{dt}\right) = 0 $$
which in turn, is a lorentz-covariant version of $$ \partial_{\mu}\left(\rho_0\dfrac{dx^{\mu}}{dt}\right) = 0. $$
I can understand the physical intuition behind the second one. An observer who looks at a given volume element and sees matter moving at speed $v$ will perceive a lorentz contracted piece of matter with density equal to $$ \dfrac{\rho_0}{\sqrt{1-\tfrac{v^2}{c^2}}}. $$
However, I don't see the motivation for the first equation (apart from imposing general covariance on the second one). The matter density there seems equal to $$ \dfrac{\rho_0c}{\sqrt{\dfrac{dx^{\mu}}{dt}g_{\mu\nu}\dfrac{dx^{\nu}}{dt}}}\sqrt{-g} $$
but I am at a loss as to why.