With the growth of consumers’ awareness towards food quality and sustainability, we are witnessing an increased demand for tools capable of detecting food adulteration. In this framework, vibrational spectroscopy allows for the rapid collection of vast amount of highly informative data to be used in food authenticity studies. This paper introduces a sparse partial membership model for food adulterant identification using spectrometry data, which are high-dimensional and characterized by complex relations among the variables. The proposal not only enables the identification of adulterated samples but also detects the percentage of adulterant while determining which spectral regions are more impacted by it. This could lead to richer chemical insights and to the development of faster portable instrument to collect data to be subsequentially used in food authenticity studies.