Recording Levels

S

stewartwarnock

New member
Hi all,

Still fairly new to this whole thing.

I was wondering what ideal db levels are for recording vocals at? What are normal levels usually? It seems as though right now i'm recording vox at an input level of -15db.

If anyone has input into this, it'd be greatly appreciated! God bless.
 
Assigning a dBfs value for recording vocals is meaningless on its own because if everything else is peaking at -3dBfs, it will not work. In other words, it's a relative exercise. Follow the link Massive posted and read about gain staging.

Cheers :)
 
I try and get peaks to be quite near 0dB without clipping usually around -3dBFS sometimes -0.5dBFS, the main reason is that when you turn up a recording that the loudest peak is at -20dBFS you are also turning up the noise floor by 20dB. A lot of people get annoyed at me though for doing this, but none of recordings have clipped yet from being very careful when monitoring the input gain and my recordings also sound cleaner as a result.
 
Where is the noise floor? A fairly noisy digital system should still be running a lot quieter than a tape machine. Normal levels going to tape are at or around line level.

If all the peaks end up quite close to 0 dB on a recording with multiple tracks, how do you get all those tracks to keep from going WAY over full scale on the master buss at mix time?

There is no standard for line level on a DA converter. Using a line level of +4 dBv RMS as an example going in, this could equal anywhere from -12 dBfs on a very old 16 bit machine to -22 dBfs on a present day pro quality 24 bit system. As a ballpark guesstimate, a lot of systems are possibly running in the -16 to -18 dBfs range. Using -18 as an example, this gives 18 dB headroom before the digital signal hits the ceiling and would have your preamp on the way in running at around line level, with a similar amount of headroom designed into it (hopefully). Peaks at -3 to -0.5 is around line level times a factor of 2 or 3. On the digital side a signal like this could be in range of having intersample peaks getting distorted without having actual samples trip the overload indicators. Or if the source material goes through an unexpected increase in volume, you clip because there's no headroom. What's even worse is that you need to overdrive the preamp to get there. It makes things sound crunchy, small and difficult to process.

How do you get rid of that crunchy sound?
 
With 16- or 24-bit digital recording there's no need to push levels up to the point of clipping, but there's no need to keep levels softer than just below clipping either. In other words, it just doesn't matter. Even budget converters have very low distortion over their entire operating range.

Gain-staging does matter, but that's not difficult to control. Just avoid running the preamp gain so low or so high that you have to raise or lower it substantially elsewhere. The files I recorded for this article let you hear for yourself that record levels don't matter as long as they're reasonable and you avoid clipping:

The Truth About Record Levels

--Ethan
 
Btyre2013 said:
I try and get peaks to be quite near 0dB without clipping usually around -3dBFS sometimes -0.5dBFS, the main reason is that when you turn up a recording that the loudest peak is at -20dBFS you are also turning up the noise floor by 20dB. A lot of people get annoyed at me though for doing this, but none of recordings have clipped yet from being very careful when monitoring the input gain and my recordings also sound cleaner as a result.
Click to expand...

That makes no sense. When you turn down the gain, the noise floor goes down as well.

Unless you're talking about the noise floor of your gear. Believe me, that's going to be much quieter than the room itself. The room has the highest noise floor in your whole chain, and that's directly affected by gain levels.

So still, your logic makes little sense.
 
The link that Massive has provided is what you should look at, 2 years ago John pointed me towards this post and I have never looked back. Read the article and do what it says, It will make a huge difference to your recordings and mixes.
 
The higher the level the more bits you are using. That means you have more audio information and a bigger dynamic range. In this mindset, you wanna get those peaks around 0.

But at the same time, all your tracks could be at -30 and you could just crank the master... you get less dynamic range, but the song will be the same volume as something tracked at -10 or even 0.

So, I aim for an RMS of -20 to -10 and a peak between -6 and 0
 
AlecBeretz said:
The higher the level the more bits you are using. That means you have more audio information and a bigger dynamic range. In this mindset, you wanna get those peaks around 0.
Click to expand...

It's a tricky mindset to follow for me. If there's an advantage to using more bits I can select 24 instead of 16. It's not a gain staging issue. Getting a clean signal with good signal to noise ratio doesn't require pushing levels to the point where there's no headroom. Depending on the crest factor of the source signal, getting those peaks around 0 dBfs by running hot levels in tracking can require you to overdrive the preamp which will impart distortion. As an added bonus, if I have to attenuate such a signal by 12 dB to make it fit in the mix anyway, the imaginary dynamic range benefit will be removed but the distortion won't. Tracking such a signal with the peaks 12 dB lower seems like a reasonable alternative.


AlecBeretz said:
But at the same time, all your tracks could be at -30 and you could just crank the master... you get less dynamic range, but the song will be the same volume as something tracked at -10 or even 0.
Click to expand...

If you're using cheap analog front end components with small power supplies in 24 bit format, that approach can still yield a dynamic range and signal to noise ratio in excess of what analog tape is capable of. It will also be a huge help in preventing the front end from adding "artifacts". What pretty much all of the mastering engineers have posted about this subject boils down to clean tracks with a ton of headroom and low distortion end up being able to handle post processing a lot better without breaking. This includes a final smash through a limter to boost RMS to unreasonable levels. If and when that happens, any discussion about dynamic range seems pointless to me.

There is a sweet spot for tracking levels. Where the level ends up being once it's in the digital domain isn't hugely important. Avoiding crunchy artifacts on the way in is.

AlecBeretz said:
So, I aim for an RMS of -20 to -10 and a peak between -6 and 0
Click to expand...

If it meets your needs that's good.

I look at the preamp and the interface, not the daw on the way in. Yellow is the new red. I aim for lots of green and no yellow. Makes it difficult for anything except maybe percussion to come anywhere close to the -6 to 0 range. If there ends up being a small amount of yellow by accident, that's okay too because the system is running with enough headroom to allow for it.
 
AlecBeretz said:
The higher the level the more bits you are using. That means you have more audio information and a bigger dynamic range. In this mindset, you wanna get those peaks around 0.
Click to expand...
While this does make sense from a digital noise floor perspective, the digital noise floor isn't the thing to worry about, the analog noise floor is. 24 bit noise floor is down around -144dbfs. The noise floor of anything you could possibly record will struggle to have a noise floor of -110dbfs. That being the case (in this oversimplified example) you could theoretically have peaks hitting -34dbfs and still capture all 110db of dynamic range.

The dynamic range of the system doesn't matter unless it is smaller than the dynamic range of what you are recording. With 24 bits, that just isn't possible in the real world.
 
I prefer -12 but I'll allow it to go above that on certain vocal lines or guitar riffs. The thing is, you don't want to record loud or it'll clip your master a lot of times. And if you're new to this, you won't know how to get rid of that. What you want to do is record everything at -12 or lower. Then, you see your master output fader over there? When you're done with the project, bring the levels up to 0.0db without any clipping by using a limiter and a compressor. Now this is different than compressing a single track, your'e compressing the whole song, so you don't want much of it, but since the limiter brought the volume up to 0.0db and it'll be just as loud as any other song you hear, that limiter will force certain dynamics down and make them not sound good. Unless they are already smoothly compressed a little and then when it's limited, it works out better.

You can use a limiter or use a compressor and add a good bit of gain back to make it louder on a single track and not the output also. But usually when on the single track you just want to compress to get the dynamics right, the goal isn't volume.

That is one of the reason's we don't want to record too low or too loud. There are other reasons to do with headroom.
 
I also want to add to that it depends on what it is. Acoustic guitar and vocals ONLY, I do it at -12db. If I'm doing a heavy metal song I may go down to -18 or even lower. You also don't want your source, like your guitar pre amp, or mic pre amp, if it's external to be clipping. A lot of people just look at the DAW and see where the level is. Some mic pre amps if not set right can clip even though the DAW shows it at -12db. I'm not exactly sure why that is, I'm not a professional when it comes to that. I'm just aware of it and I have to be careful because I've messed up someone's vocals bad before because I ran the pre amp too hot yet the levels were not clipping and I was recording at unity.
 
Technically, the -18dbfs thing should be the average level, not peak level. For vocals, I have the singer sing a sustained, loud note and set the level of that to -18dbfs. Everything else takes care of itself. You do need to watch out for clipping with vocalists that aren't as controlled.
 
You must log in or register to reply here.
Back
Top