YT compression is half the story, my main concern is with the source. The audio mode used to play the game when being recorded is what matters most. Let's say one game is running with 2 channel LPCM, while the other game is in DTS 5.1, then the in game sound mode may offer mixing ( dynamic range, audio level adjustment to music, sound effect, speech etc ), these differences can give overall different peak audio output, dynamic range and mixing. I gave example with GT6, the game has 3 different dynamic range level, several audio modes like stereo, DD and DTS that each gives different sound level.
Here's the tricky part, game with low dynamic range often tend to sound louder/noisier as the differences between low to high are narrowed, while high dynamic range tend to sound quieter/subdued.
I don't know if YT does the audio conversion or the uploader have to do it themselves ( I never uploaded video with multichannel audio in DTS or DD format ), downmixing multichannel audio can also be a variable ( depending on the software matrix adjustments )
Anyway, I still don't like GTS sound quality, the sampling rate is one of the offender IMO., not just the quality of the recording or the amount of details processed. If you ever played MGS IV, that game has one of the best audio sampling rate, clarity even on DD 5.1