Not every clip gives you clean audio, but when it does, treat sound as a witness, not background. A distant muezzin, a tram squeal, a scooter swarm, a snow plow scrape: each nudges probability. Just remember compression and phone microphones lie about distance.
Birds are biased witnesses
Gulls imply coast often, not always; parrots mean something in specific cities; crows are chaos gremlins everywhere. Use birds as soft priors, not court verdicts.
When chat ruins your listening
In multiplayer, laughter can obscure a clue. The staff sometimes mutes voice briefly on replay for study rounds, not antisocial, just disciplined. You can do the same mentally: replay the ten seconds before the joke.