Situation is this. I recorded a bunch of old casette tapes (peaks around -6 to -9dB) and I want to apply equalizer to them to correct the lack of bass. The problem is that I probably need fairly large adjustments, so I will get clipping if I just apply the equalizer directly. of course I can use the “amplify” or “normalize” effect to take down the volume first, then apply equalizer, and then amp up the volume again for a finished master - but is there a smart way of calculating how much I need to down-amplify each recording in order to fit the equalizer settings without clipping?
The point is that I just want to avoid “crushing” the details of the recording too many/drastic amplify actions - so is there a smarter way to do this other than just guessing (and having to waste a lost of time re-doing the filters with different settings because i’m getting clipping)? With 16 files and about 2 minutes pr. operation (normalize and equalizer) it can end up being a real chore to do it via the “guesstimate” aproach.
Hopefully one of the veterans here can help me out
If you are using the default “32 bit float” format in the audio track (see the info panel on the left end of the track), then you may amplify the track up and down as often as you like without doing any damage (assuming that you don’t try amplifying by hundreds of dB )
With a bit of practice you will be able to guess fairly accurately how much the waveform will be boosted or cut. Note that Equalization can go down as well as up. If you turn down the levels of all frequencies except for bass, then that is equivalent to turning up the bass and reducing the volume.
I am using 32bit float yes.
If I’m not mistaken there will always be some level of detail loss for an operation like this, especially the more drastic the change of levels you do up and down - but I am assuming what you mean is that 32bit float is high enough precision that it doesn’t really matter practice in normal usage (ie. not hundreds of extreme level changes). I understand that. Especially given that the source material is poor (15-20 year old casette tapes) I know it’s not REALLY an issue. I’m just a little ADD when it comes to this sort of stuff being done in an “optimal” fasion whenever it is possible. If you cut the volume in half and then amp it up again digitally you do effectively lose half the precision of the samples being used right? It’s possible I’ve misunderstood something…
I’d still like to know if there is a better way to do it - but your answer is very reasonable.
Quite so. In fact you can perform literally hundreds of amplification operations without any significant losses. 32 bit float format operations are accurate to within something like 0.000000000001% and the dynamic range of 32 bit format is somewhere in the region of 1000dB. You can go pretty extreme with amplifying 32 bit float format and the losses are barely measurable and certainly not audible.
No, not with 32 bit float format. The “float” part means that it can handle fractions. If you exactly halve the amplitude, and then exactly double the amplitude, the result is identical to the original (the mathematics that make this possible is rather complicated, but it can be demonstrated relatively easily).
32 bit float format is vastly more accurate than the highest quality audio hardware. The errors produced by this type of operation are far too small to register on the highest quality audio hardware (and that is why it is the default format).
Alright then - my understanding was flawed. Thanks for clearing that up. I won’t waste my time with minimizing use of amplification anymore
WAIT! It’s still good practice to keep your “final output” below 0dB.
Before exporting, I recommend you run the Amplify effect to bring your peaks down to 0dB.
Although floating-point won’t clip, your DAC (analog-to-digital converter) can clip if you go over 0dB. If you export to 16-bit or 24-bit WAV, or if you make an audio CD, the audio will be clipped.