Dumb Gain Staging Question - Soft Tracks/Channels

[Sundown]

Hey all,

 

I understand a decent amount about gain staging within a DAW (or in a setup), but there are still things I have to learn or improve upon.

 

If I have an individual channel/track that I *know* is going to be low in the mix, it still makes sense to me to record it as close to the optimum input level as I can (which I typically treat as peaks between -18dBFS to -12dBFS, and an average of -18dBFS or 0VU). 

 

Mind you, when I go to mix, I’m going to pull the gain down (or the fader), but with this approach, I’m still limiting the amount of noise floor in the signal. It doesn’t make sense that I would capture the signal low on the meter, or it doesn’t make sense to me at least.

 

Keep in mind I’m a studio keyboard player so typically tracks will get recorded as MIDI first, and then I’ll record them as audio after the fact. I also record at 44.1KHz/32-bit float with RME 24-bit converters.

 

Does this make sense? Or would you record softer sounds closer to their true level in the final mix?

 

Todd

[Anderton]

By recording at a higher level, you minimize distortion (better bit resolution) and noise (higher level compared to the noise floor). 

[Stokely]

I'd always record at optimal level for each track, but then I am old school and learned when tape hiss was the nemesis    

Easy enough to turn things down without any downside.  Even if it was some throwaway background noise for an intro where you'd never want it loud, I'd still try to get a good recording.  If the fader goes down so far as to make it hard to mix you could always turn down the gain further up in the chain.   I saw an interesting vid where someone went through each track kind of "normalizing" gains (almost the opposite of what you were mentioning!) before any mixing was done...even region by region if there were jumps in say a vocal track.

[Anderton]

24 minutes ago, Stokely said:

even region by region if there were jumps in say a vocal track.

 

I do this all the time with vocals. Not only does it give the vocals more clarity, it reduces the need to use compression so you don't get dynamics processing-related artifacts. That, and other normalization techniques, are mentioned in the article Why Is Normalize a Dity Word?

[Sundown]

Thanks again guys - Huge help.

 

Todd

[Sundown]

7 hours ago, Stokely said:

I saw an interesting vid where someone went through each track kind of "normalizing" gains (almost the opposite of what you were mentioning!) before any mixing was done...even region by region if there were jumps in say a vocal track.

 

Yep - I’ve seen that too. I guess the argument is that faders are more sensitive at unity, so turning down gain pre-fader preserves that resolution. 

[Stokely]

16 hours ago, Anderton said:

 

I do this all the time with vocals. Not only does it give the vocals more clarity, it reduces the need to use compression so you don't get dynamics processing-related artifacts. That, and other normalization techniques, are mentioned in the article Why Is Normalize a Dity Word?

Compression is kind of where I'd see this being kind of important actually....with the premise being that compression acts on the signal post-gain but pre-fader (hopefully that is correct as far as the insert point in signal flow in Logic!)   I use compressors a lot but I think if there were wild jumps you'd as you say get big swings in sound quality as it squashes more in certain places by going way over (or staying under) the threshold.

Of course if you have time and control over a track to re-record another take that's what I do, but if it's not your vocal or it's a tough part that you otherwise nailed, I don't see why bumping the gain would be a dirty thing    (noise being the thing to check for)

[Anderton]

Manual normalization is more time-consuming, compressors save time. But I hardly ever use compressors on vocals, I think it takes away some of the intimacy. I've done two songs recently where at one point, there was only voice and one other instrument. Any compression would be audible. I sometimes add a few dB of limiting, which is pretty much transparent. I also use gain envelopes a lot to bring down "rogue peaks." Then normalization gives a higher average level, again without compression.

[KuruPrionz]

1 hour ago, Anderton said:

Manual normalization is more time-consuming, compressors save time. But I hardly ever use compressors on vocals, I think it takes away some of the intimacy. I've done two songs recently where at one point, there was only voice and one other instrument. Any compression would be audible. I sometimes add a few dB of limiting, which is pretty much transparent. I also use gain envelopes a lot to bring down "rogue peaks." Then normalization gives a higher average level, again without compression.

I've been learning to lean in when singing softer and pull back when I'm getting loud. That is helping keep things more even but it takes both practice and keeping an eye on your channel meter so you know what your volume is going to be.