I'm producing an animation and I came to a frustrating dilemma with the sound normalization.
The recommended normalization for dialogue, on YouTube, is apparently between -6 db and -15 db. Due to many effects added to the voices of the characters, some dialogue lines endup being much louder than others, even after they are all normalized at -6 db peak amplitude.
To solve this problem I decided to normalize them at different peaks amplitude. Some at -8, some at -2, while most of them are at the recommended -6 db. I have to point out that none of the lines ever reach a level louder than -7 in real time. In other words, the actual volume level (to the ear) is never louder than -7, even those normalized at -2. My concern is that "YouTube will turn down the volume of the video if the audio is normalized lower than at -6" REGARDLESS if the actual volume levels of those normalized at lower than -6, are actually quieter to the ear. I'm worried that the system which corrects the audio on YouTube may not only be guided by the volume level, but by some other indicators, which may show that the normalization was lower than at -6.
Question: Should I normalize the lines at whichever peak amplitude i have to - to reach a level satisfactory for the animation and just make sure the actual loudness doesn't go over -6db, even if it means normalizing it at -2?
This issue is the last thing I have to solve before the animation is finally ready after months of work. Hand on my heart, I would be extremely grateful for any help.