Efficiency in bioprocess development is often limited by sampling frequency, working times, sample treatment, and analysis. In-line Raman spectroscopy enables the real-time monitoring of relevant substance concentrations directly inside the reactor, thereby rendering most time-consuming and error-prone analysis procedures redundant. 

Here, two industrially relevant fermentations for pullulan and triacylglycerol production were characterized using offline measurements to facilitate future in-line applications. An automated data processing pipeline was developed, and correction strategies for the inner filter effect were investigated using a combination of broths with and without yeast cells (WiC and WoC). Spectral regions for substrate, metabolite, and cell-dry weight (CDW) identification and quantification were defined. First preliminary multivariate Partial Least Square Regression (PLSR) models were developed, yielding Root Mean Square Errors of Prediction (RMSEP) between 1.4 g/L (CDW) and 6.9 g/L (pullulan).