However, that's not a particularly scientific approach, and requires a lot of trial and error to get the right balance. Decide which one is softest, and reduce the levels of the other one to match it. One way to do that is to normalize all the cuts in the album to something under 0 True Peak initially. If you're doing an album project, ultimately it's up to your ears to decide if the levels are balanced.
True Peak meter readings take this into account. A signal that registers as '0' on a meter that's reading sample values could actually be creating a signal that's 3 dB over the available headroom after conversion.so there's distortion even though the meter says there isn't. If you normalize to 0, you run the risk of intersample distortion due to the D/A converter's smoothing process. If you plan to add limiting or compression, turning up the processor's input level, or lowering its threshold, is functionally equivalent to normalizing the input signal to a higher value. It's no different from turning up the level. Normalization doesn't affect the sound or the dynamics.
