Lossless audio representations of molecular information

A common method to rapidly communicate molecular information is by representing this information in the visual domain in a process called visualization, for example via molecular drawings. The analog of this in the sonic domain is called sonification, an approach rarely used to communicate molecular information. This project will involve the creation of representations of molecular information (e.g. spectra, atomic connectivity, ...) in the form of sound. As an additional constraint, these representations will be lossless, in other words, the molecular information should be fully recoverable from the audio representation.

- Familiarization with the state of the art in molecular sonification as well as sonification in science in general. Critical assessment of shortcomings of current methods.
- Implementation of lossless audio representations using general purpose software (such as Python, Matlab, ...) or dedicated audio processing frameworks (such as LibRosa, CSound, PureData, SuperCollider and others). This implementation can be inspired by existing approaches such as SAMPLES sonification of molecular graphs (Mahjour, 2023)
- Assessment of the created audio representation: how salient is the signal, which aspects of the molecular information can the representation communicate uniquely, can the representation be used for accessibility purposes?
- Round-trip benchmarking to quantify the recoverability of molecular information