Detection and simultaneous monitoring of multiple trace gases is vital in scientific and industrial processes. Here, we use coherent control in quartz-enhanced photoacoustic spectroscopy (COCO-QEPAS) with an in-situ learning method for rapid fingerprinting of trace gases to identify and monitor arbitrary gases at very low concentrations, without prior knowledge of gas composition. We validate this on various mixtures, including CH4/C2H2/C2H4/C2H6/NO/NH3. To this end, we demonstrate real-time analysis of mixtures containing up to four trace gases at ppm-level, monitoring changes in seconds using linear regression. The scalability of simultaneously distinguishable gases is straightforward. Furthermore, we expand fingerprinting to 10 ppm with a detection limit of 180 ppb CH4, and apply empirical mode decomposition as an adaptive, data-driven filtering method to recover characteristic spectral features at the noise floor. For quantitative analysis in the ppb regime, we employ principal component regression as a calibration model that ex-ploits correlations across the full spectrum. Consequently, our method offers significant potential for sensing applications where speed, accuracy, and simplicity are critical. The data is stored in individual subfolders, you can just run the python script in the respective folders with the specified version. This should enable you to reproduce the data.