a limiter is not used to raise levels (unless you are using makeup gain to compensate for other issues). A limiter is designed to assert a maximum level that our channel(s) can achieve by folding any signal that moves above that level back to that level
Been thinking about this overnight.
My first response was heavy on mathematics and other theoretical considerations, so did not post it; this one has much less mathematics in it, though some may disagree.
Effectively, your issues surround the difference between perception of sound levels and the recording/creation of sound levels in a daw.
I.E. normal background sound is perceived as being at 60db, quiet sound is 20db (middle of the night in the country), loud sound is 105db (full orchestral tutti at fff or louder), threshold of pain is 130db, 0db is the threshold of hearing, orchestral ppp is 40db
Description | nominal db level | difference from Normal | difference as a ratio |
---|
threshold of hearing | 0db | -60db | 1:1 000 000 |
orchestral ppp | 40db | -20db | 1:10 000 |
quiet sound | 20db | -40db | 1:100 |
normal | 60db | 0db | 1:1 |
orchestral fff | 105db | 45db | 31 623:1 |
threshold of pain | 130db | 70db | 10 million:1 |
From the table we can see that the range of a full crescendo from ppp to fff covers 65db (a ratio of 3.16 million:1) - this is much less than the dynamic range of typical cd audio 16 bit 44.1kHz (96db). Our daws, on the other hand, can handle at least 124db dynamic range (ca. 20.6 bits - ratio of 2.5 trillion:1 - this is actually the limitation of most dacs rather than the mathematics in the audio engine), almost the full range of 24 bit audio - 144db (ratio of 251 trillion:1).
If you try to create a crescendo (increasing the level of the signal continuously over time) that exceeds 65db, you are probably going to run into issues of representation, volume levels and other psycho-acoustic issues