Vibrational spectroscopy methods - near-infrared (NIR) spectroscopy and Raman spectroscopy - have become valuable analytical tools in various fields in both science and industry. This is due to their ability to provide unique information about molecular structure, the possibility of non-contact measurement, the possibility of real-time analysis, the ability of non-invasive analysis of complex systems, the minimal interference with water, and the complementarity of application with other techniques. However, Raman spectra contain additional negative influences beyond the Raman spectrum, such as fluorescent background, noise, and cosmic rays, which must be removed to ensure measurement reliability. To overcome the challenges of interpreting complex spectra obtained by Raman and IR spectroscopy and to provide deeper insight into the data, various methods of data preprocessing must be applied. This final thesis provides a theoretical overview and analysis of the application of data preprocessing methods in Raman and IR spectroscopy, such as baseline correction, data normalization, noise reduction, outlier detection and extraction of essential characteristics from the spectrum. In this thesis, in addition to a method overview and examples from the literature, the influence of the selection of different parameter and data preprocessing sequences on the overall quality and reliability of the processed spectral data is investigated using laboratory data from the crystallization process of the active ingredients.