This question already has an answer here:
I am aware that we can predict the colours of compounds, on the basis of their electron transition energies, and many other physical properties (on the basis of other theories.)
Smell works in an inherently different manner, as the compound binds with smell receptors in our noses, and we smell a distinctly different smell for many compounds, quite unlike colour, or crystalline/amorphous nature, etc. It's also highly subjective, varying slightly from person to person. But, so does observed colour...
However, I don't see any way of predicting the smell of a random substance with any technique, except when we know what it is (e.g. hydrogen sulphide smells like rotten eggs, ammonia like rotten fish, benzaldehyde like almonds, etc.)
So, I guess my actual question is, is smell an actual physical property affected by the chemistry of a compound, or is it highly random? Are there any theories predicting the smell of compounds on the basis of their structures, functional groups, etc?