You want to reduce transients but don't lose dynamic range. That varies from sound to sound but you generally want to reduce peak and increase average loudness. You can do that with a few different processes, including saturation, compression and limiting.

Then for resonances and general frequency spectrum you normally want to reduce frequencies around 2k or 4k hz in case they are too harsh. And depending on proximity of the sound you may want to increase or roll off higher >8khz frequencies. Then get rid of resonances on the mid range without loosing weight of the sound, so don't kill the low mid frequencies completely.

Then all that combined you probably want to have more dynamics on the low mids and keep the highs more compressed so multiband processing may be a good idea. But you may want to compress highs less, etc. depends on the sound.

Speaking of which, sound selection is the most important so if you are in the middle of processing don't be afraid to change the source and try different variations.

Finally make sure to monitor the sounds on different headphones or devices to hear how they will sound for the player.