What is wrong with AI sounds in a game
AI audio is made to sound good on its own, not next to the other sounds in your game. Sound effects often start with a quarter or half a second of silence, so the hit plays after the hit. Loudness jumps from one file to the next because each one was made on its own. Some files carry a DC offset, a small shift of the whole wave away from zero that pops when the sound starts and wastes headroom. Many are fake stereo, two identical channels that double the size of the file. Peaks can go over full scale and crackle, and a music track that ends anywhere clicks every time it loops.
What this tool fixes
Silence. The quiet part at the start and the end is cut, with a few milliseconds kept so the attack stays sharp. Loudness. Every file is set to the same loudness in LUFS, the measure streaming services and game audio use, so a pack of effects sits at one level. Or set them all to the same peak. Peaks. A limiter holds the loudest moments under the ceiling, so raising the level never clips. DC offset is removed and fake stereo becomes mono. Fades stop clicks at the ends. Loop. For music and ambience the end is crossfaded into the start, so the track can play forever without a seam. Press Hear the loop point to listen to the join.
Which loudness to pick
The presets set sensible starting points: music around -16 LUFS, sound effects around -16 LUFS with a fast fade, UI blips a little quieter, ambience lower so it sits under everything, and voice around -18 LUFS. Mixing in your engine is still the final word, but starting from one level saves a lot of guessing.
Is my audio uploaded?
No. Everything happens in your browser. The samples are made in code with the problems AI audio often has: long silence, a low level, a DC offset and fake stereo.