Food safety and quality is of worldwide concern and for meat the authentication of different species is a frequent issue with many implications including economic, religious, ethical, and health issues. Common analytical methods for meat species authentication are mostly labor-intensive, time-consuming and expensive. Optical techniques are a promising alternative enabling rapid and non-invasive in-situ analysis. This study extends our previous investigations to the analysis of frozen-thawed meat and meat juice using Shifted Excitation Raman Difference Spectroscopy (SERDS) applying two miniaturized SERDS probes operating at 783 nm (110 mW optical power, 0.5 nm spectral shift) and 671 nm (40 mW optical power, 0.7 nm spectral shift) that are fiber-optically coupled to compact spectrometers. Specimens comprise pork, beef, chicken and turkey and for each species 12 fresh meat slices were frozen at -18 °C for 7 days. After thawing each slice was measured at 15 different spots while for the meat juice 5 drops originating from each slice were analyzed recording 10 spectra with integration times of 10 seconds each. Partial least squares discriminant analysis models using 4 latent variables showed a clear distinction between individual species for meat (Sensitivity < 94 %, Specificity < 92 %) and meat juice (Sensitivity < 97 %, Specificity < 98 %). The classification is based on variations in myoglobin content and complex differences in protein Raman bands. The results underline the large potential of SERDS for rapid and non-invasive in-situ meat authentication paving the way for future applications at selected points along the process chain.