The two dimensional J-resolved spectroscopy technique is capable of resolving many metabolites in vivo from a volume of interest. However, the spectral resolution along the indirect dimension is generally very poor in these acquisitions. One solution is to apply a covariance transformation along the indirect dimension to yield a resulting Covariance J-resolved (CovJ) spectrum with high spectral resolution. While spectral resolution is enhanced, currently there are no methods available to fit the non-linear aspects of the covariance reconstruction. Here, we have developed a non-linear fitting algorithm capable of yielding Glutamate, Glutamine, GABA, and Glutathione concentrations in vivo using CovJ spectra.