When capturing audio in games, the sound levels vary. Sometimes something is said loudly, sometimes quietly in a distance.
When actually learning from the deck it can be a jarring contrast to flip from something very quiet to something very loud.

So far when building decks with Genshin Impact, I recorded in parallel in Audacity and applied the Normalization effect to normalize the audio to -3db so everything in my deck is the same loudness. (technically not really, since audacity's naive normalization ignores human hearing as expressed by the industry standard LUFS measurement, so a bright female voice is still louder than a dark male voice)

It would be really awesome, if Game2Text could normalize the audio to some equal measure, during the conversion step from .wav to .mp3. That would improve the quality of Anki Decks generated by Game2Text.