The spectroscopic quantification of mixture compositions usually requires pure compounds and mixtures of known composition for calibration. Since they are not always available, methods to fill such gaps have evolved, which are, however, not generally applicable. Therefore, calibration can be extremely challenging, especially when multiple instable species, e.g. intermediates, exist in a system. This study presents a new calibration approach that uses ab initio Molecular Dynamics (AIMD)-simulated spectra as to set up and calibrate models for the physics-based spectral analysis method Indirect Hard Modeling (IHM). To demonstrate our approach called AIMD-IHM, we analyze Raman spectra of ternary hydrogen-bonding mixtures of acetone, methanol, and ethanol. The derived AIMD-IHM pure-component models and calibration coefficients are in good agreement with conventionally generated experimental results. The method yields compositions with prediction errors of less than 5% without any experimental calibration input. Our approach can be extended, in principle, to IR and NMR spectroscopy and allows for the analysis of systems that were hitherto inaccessible to quantitative spectroscopic analysis.