Because scent-languages would presumably be composed of several dozen (if not several thousand) scent components. Detection by artificial means is a hard problem, even in limited and highly lucrative domains (truffles, explosives, etc), which is why we still mostly use dogs/animals. Synthesis would require manufacturing stocks of all the various smell components, then aerosolizing them in the right combinations.

Audio, by contrast, has one variable--air pressure--to control with respect to time.