Electronic noses (e-nose) and optical noses (o-nose) are two emerging approaches for the development of artificial olfactory systems for flavor and smell evaluation. The current work leverages the unique optical properties of semiconducting single-wall carbon nanotubes (SWCNTs) to develop a prototype of a novel paper-based near-infrared optical nose (NIRON). We have drop-dried an array of SWCNTs encapsulated with a wide variety of peptides on a paper substrate and continuously imaged the emitted SWCNTs fluorescence using a CMOS camera. Odors and different volatile molecules were passed above the array in a flow chamber, resulting in unique modulation patterns of the SWCNT photoluminescence (PL). Quartz crystal microbalance (QCM) measurements performed in parallel confirmed the direct binding between the vapor molecules and the peptide-SWCNTs. PL levels measured before and during exposure demonstrate distinct responses to the four tested alcoholic vapors (ethanol, methanol, propanol, and isopropanol). In addition, machine learning tools directly applied to the fluorescence images allow us to distinguish between the aromas of red wine, beer, and vodka. Further, we show that the developed sensor can detect limonene, undecanal, and geraniol vapors, and differentiate between their smells utilizing the PL response pattern. This novel paper-based optical biosensor provides data in real-time, and is recoverable and suitable for working at room temperature and in a wide range of humidity levels. This platform opens new avenues for real-time sensing of volatile chemical compounds, odors, and flavors.