Synthetic dataset from the manuscript "Deep learning recognition and analysis of Volatile Organic Compounds based on experimental and synthetic infrared absorption spectra"

Synthetic dataset of IR absorption spectra of volatile organic compounds (VOCs) generated by the conditional variational autoencoder described in the manuscript 'Deep learning recognition and analysis of Volatile Organic Compounds based on experimental and synthetic infrared absorption spectra'.
 
--------------------------------------------------------------------------------------------------------

Each folder contains multiple .npy files each containing 10 generated spectra associated to a fixed concentration in parts per milions [ppm]. The concentration is reported in the file name:
{class_name}_PPM{concentration in ppm}.npy

The concentrations span the range of values of the experimental dataset, generated with a step of 1 ppm.

The dimension of each file is (10, 622), where 10 represents the different spectra and 622 is the number of channels corresponding to the range 700 - 1300 cm^{-1}.

Directory tree structure:

- Air
- Acetone, 82 concentrations from 5 to 86 ppm
- Benzene, 66 concentrations from 18 to 83 ppm
- Ethanol, 40 concentrations from 9 to 48 ppm
- Isopropanol, 92 concentrations from 1 to 92 ppm
- m-Xylene, 63 concentrations from 15 to 77 ppm
- o-Xylene, 49 concentrations from 33 to 81 ppm
- p-Xylene, 20 concentrations from 40 to 59 ppm
- Styrene, 81 concentrations from 1 to 81 ppm
- Toluene, 58 concentrations from 26 to 83 ppm