Perceived Volume

Monkey Allen

Monkey Allen

Fork and spoon operator
I've done measurements of pro songs and some of my songs in terms of RMS value...the average volume/ level of the song. I've measured some of my songs to be -12db RMS for example and then measured some pro songs that are -15db RMS....and the pro songs sound effortlessly louder than mine. I'm not a loud guy...not looking to make my stuff ear splittingly loud. Not concerned about being loud. I'm just interested in this apparent fact that the pro song can measure a lower RMS and yet still sound louder.

So I was wondering if anyone can explain this technically. Is it to do with the instrumentation and the arrangement? Is it to do with the clarity of the up front parts of the pro song that make it so much clearer and louder?

Another reason I ask is because a song I just finished today sounds louder than some of the older stuff I mixed...but has a lower RMS. I usually master to K14...pushing levels until I'm sometimes hitting the red in that K14 measuring method. However, the song I just finished...I was pushing it and pushing it trying to get it into the red...but it just got too loud. In the end I let it just hit into the yellow and it was loud enough. In the end it had a lower RMS than the old recent songs I have done...but its volume was still right there and even more robust...at a lower RMS.

I was wondering if maybe that is a sign that my mixes are improving. I dunno.

I've heard people here and there shooting for RMS levels of -8db and stuff like that. Which seems really loud. My stuff usually falls around -13, -14. This latest one is -15.
 
Do you mean LUFS instead of RMS? They are related, but the "loudness" of a song is measured in LUFS. You should be dropping your mix (and things you want to compare to) into something like YouLean's loudness app.

Our ears/brain hear different frequencies differently, and LUFS attempts to [somehow] take that into account, as well as adjust for spoken word and other stuff (I don't know what all it does). In any case, if it measures the same LUFS, your sense of "volume" at a normal listening level should be sort of the same. And, conversely, if the LUFS of two different tracks is significantly different, 3dB or more, you should definitely perceive the higher value as louder. If not, then I'd wonder if there's something in your monitoring system or space that is skewing thing.
 
Right. My song's LUFS is -14. The pro song is -17 but the pro one sounds louder and fuller and richer...despite being -17 LUFS compared to my song's -14 LUFS. Sounds like if I play them back to back, my song first then the pro one...sounds like right after my song, someone reaches for the volume knob and cranks it 4 or 5 extra notches up on the old stereo.
 
Can you put at least a 30s clip of both in MP3s, i.e., that measures the same LUFS and is perceived the same way, and attach to this thread. I’m curious what might be going on.
 
OK. This is complex but not at the same time.

First, Fletcher- Munson curve. Human ears and brain register sound best at the mid frequencies where human voices live.

Second, low frequencies will take more "power" to reproduce which affects the usable dynamic range of a given song.

Third, compressors and limiters are going to hit the highest energy sounds first before they get any where near the RMS stuff, IOW transient peaks and low frequencies tend to "drive" comp/limiters (unless you high pass the detector, which will only change how it reacts to lows). So mix bus comp/limit RMS is really a function of peak value, the more dynamic it is going in the less it can be pushed to be "louder" without causing issues.

So what does all this mean? The short answer is that each part of the song whether tracks or "stems" needs to have a limited dynamic range because that way, when added together, the full spectrum will be filled but contains fewer large peaks, and hopefully a better balanced low end which allows mix comp/limiting to raise the perceived volume and RMS together.

How to do this?

Limit dynamic range. The Singer doesn't need any thing below 150-250 hz or above 5 k really (on average). Every other instrument can be similarly sculpted.

Volume automation. Riding a fader while writing an automation track for everything is time consuming. But it insures that any comp/limit done afterword is going to be most effective at keeping RMS at target level without pumping or artifacts.

Parallel processing. Sounds with strong attacks and short percussive wave forms such as drums, guitar palm mutes, fingerstyle acoustic guitar, etc have huge dynamic range because of the spikes caused by the attacks and the low level of the sound after the spike. Comp/limiting will reduce the spike and can bring up the volume of the decay but it can EFF up the sound too. Sending these types of sounds to an aux track and using saturation, light distortion, compression allows you to slam that spike right down and bring that decay right up without messing with the attack of the original at all. Adding the two together on a buss gets the best RMS because it effectively limits the dynamic range.

Work with stems. Once each track is in it's best shape if you send them to aux buses for say, drums, guitars, keys, etc it is easier to find and control large dynamic swings that mess with perceived loudness and RMS. Plus once you have a good balance on each stem you can vary the balance between the sections to emphasize different parts creating a more interesting mix without crazy volume changes. Though a crazy volume change as a creative tool can also be done more easily with stems.

Use mid/side processing and automation for panning. Moving important parts to the center and away when they are no longer featured allows for a more consistent volume/RMS because of pan law. It's a lot easier to hear everything when they all have their own space.

Anyway , just some basic ideas, YMMV
 
Gtoboy said:
Limit dynamic range. The Singer doesn't need any thing below 150-250 hz or above 5 k really (on average). Every other instrument can be similarly sculpted.
Click to expand...
I disagree. There is some 'air' that is part of the body in the voice.

If I HFP LPF filter my vocals 250-5k it is ...tight.

I typed out a big response and clicked cancel....
 
If I wanted to make it louder, I would try and limit the range. Then possibly saturate it. Limit at -6db and push against the brickwall. It will technically be quieter , but sound louder . I think you can saturate any type of circuit that has a input gain and output gain. You turn the input up against turning the output down. It overdrives itself..I think that is saturation from the circuit. I dont know..
Screenshot 2022-04-09 181222.jpg

A quick fresh midi doodle. I hawkeye'd it at -14 LUFs. Does this sound appropriate in volume? Everyting looks 'green'.
Screenshot 2022-04-09 180016.jpg
Loud?


Same-ish target levels NO LIMITING. Look how much difference limiting it makes in loudness. Er listen..whatever.


The first one, limited, sounds much more exiciting.
 
Last edited:
LazerBeakShiek said:
I disagree. There is some 'air' that is part of the body in the voice.

If I HFP LPF filter my vocals 250-5k it is ...tight.

I typed out a big response and clicked cancel....
Click to expand...
The idea is not to get rid of anything that sounds good (in the mix), more of get rid of stuff that is taking energy away. Such as ac rumble or the hss of air movement that can't be heard in a full mix yet is taking up energy being reproduced.

My whole post was done at work during a break so I kept things simple as possible. When I actually implement filtering, I always start at top/bottom and move toward the center until I can hear a change, then back away until I can no longer hear that obvious difference, When doing it with a full mix going you will be surprised how far you go, bur I admit the numbers I gave are not hard and fast rules nor were they meant to be. They are values that I have found are in the areas I generally end up using in a mix context, After the basic mix I can always add back some air, some growl, things like that.

Also, many times, for both voice and played instruments I will use shelving EQ instead of filters, and manipulate the "Q" and crossover frequency to put a resonant bump in a particular spot.

Finally, with my voice and male voices in general, I am using the high pass closer to the 150 hz I originally stated, and female voices(or kids!) at the 250 hz.

I did not intend to give the definitive answer, just some thoughts and points I have found helpful, because I have used these concepts enough to know that I can easily get a mix much louder than the average commercial mix without clipping or artifacts.

Here is an experiment anyone can do with an EQ and a spectrum analyzer:

Take a recording of a single voice or instrument and add a low pass filter. No analyzer yet. Slide the filter up until the sound is unacceptably thin and back down until it isn't.
THEN put an analyzer on it. Pop the EQ in and out and see how much low frequency stuff was using up dynamic energy. Watch your peak and rms meters.

Don't throw away the baby with the bath water but do throw away that dirty bath water! Unless, and this caveat always holds: unless that's the sound you are going for. Mixing is trade offs because there is only so much real estate to work with. If you take up all of the real estate with your tool shed where is the house going to go? On top of the shed? Yeah it will fit, but it's not going to make the shed or the house particularly useful and is more of a nightmare. Again, unless that is the effect you INTEND.
 
Gtoboy said:
Here is an experiment anyone can do with an EQ and a spectrum analyzer:

Take a recording of a single voice or instrument and add a low pass filter. No analyzer yet. Slide the filter up until the sound is unacceptably thin and back down until it isn't.
THEN put an analyzer on it. Pop the EQ in and out and see how much low frequency stuff was using up dynamic energy. Watch your peak and rms meters.
Click to expand...
Or flip the EQ , to only listen to what is being cut.

Are you into heavy processing while recording? You just use a preamp->EQ ? Or do a Preamp->EQ->AGC->Compressor->limiter OR preamp->EQ->compressor->compressor?
 
I tend to record without processing except for bass guitar. About 50% of the time (depending on the arrangement/genre) I put the bass DI track through a compressor on the way in, though, oddly enough, I don't usually compress the bass amp mic. In fact, I think its only happened a couple of times.

Depending on the room I used to use high pass filters at ~60 hz to keep from recording rumble , but now that I have my dedicated recording space I don't even do that . On rare occasion I might high pass a hi-hat or cymbal mic on the way in, but as I said-rarely.

Most of my tone shaping comes at the source or through heavy processing after if I decide on something crazy
 
LazerBeakShiek said:
If I wanted to make it louder, I would try and limit the range. Then possibly saturate it. Limit at -6db and push against the brickwall. It will technically be quieter , but sound louder . I think you can saturate any type of circuit that has a input gain and output gain. You turn the input up against turning the output down. It overdrives itself..I think that is saturation from the circuit. I dont know..
View attachment 116952

A quick fresh midi doodle. I hawkeye'd it at -14 LUFs. Does this sound appropriate in volume? Everyting looks 'green'.
View attachment 116949
Loud?


Same-ish target levels NO LIMITING. Look how much difference limiting it makes in loudness. Er listen..whatever.


The first one, limited, sounds much more exiciting.
Click to expand...

So I finally got a chance to listen to these and here are my thoughts on the processing:

Because of the Fletcher Munson curve louder is ALWAYS going to sound more exciting to the human ear which is why as mixers we are supposed to volume match when going through processors(except a limiter). It is very hard for a human ear to perceive subtle differences and make an informed choice when large volume changes are involved.

So the loud version I was confused when listening. It sounds (to me) chaotic. Everything is as loud as everything else so I don't know what to listen to. I just get the sense of the same loud wave crashing against the same wall repeatedly. So I almost didn't listen to the quiet version!

However, I did listen to the quiet version and now suddenly I hear a focal point in the center, One single instrument that stands out while everything else becomes background in my headphones and I start to bob my head to the melody that is now clear. I can listen without being overwhelmed by sound. Is it "better"? No. Different, Intention is what counts. What is the end goal? How do I want the Listener to Feel/React?

As for the flipping eq , yes, if you have good subwoofers you will "hear" what you have cut out but I still feel an analyzer will give a more thorough understanding of how much dynamic space is being taken up by stuff that you just don't need in your mix.

As always, just my opinions and experience and YMMV
 
@Monkey Allen - any update? (Clips?) I did have a thought that your own mix's LUFS measure can be skewed by some things you can't actually hear, like sub-sonic noise from an A/C compressor, fridge compressor kicks in (AC in the attic and fridge downstairs! - both have happened to me). Assuming you're actually comparing like mixes/genres, which is what you should be doing, then I'd look at the frequency curves - something like iZotope's Total Balance Control is what I use, to make sure something isn't slipping in there that you're own ears, or room, might be missing or masking, but which is contributing to the loudness measure.
 
You must log in or register to reply here.
Back
Top