Effects of raising the volume

[Chris++]

Good morning, all. If anyone would be so kind, I have a question about something that probably should have occurred to me years ago. It has more to do with basic audio science than anything, I think.

I record songs using pretty standard, popular-level recording programs, such as Audacity and Nero Wave Editor. That may be irrelevant, however.
Once I'm done mixing, equalizing, and otherwise finalizing a song, I tend to maximize the overall volume (by using "Normalize," usually).

Will raising the volume tend to change any other properties of a sound?

It's kind of a silly question, since a volume increase "shouldn't" have any effect apart from making something louder, but I've occasionally been a bit surprised at how the top or bottom will seem to be exaggerated once I've maximized the level of a whole song. It all just gets a bit harsher, so I've been wondering if any other particular characteristics of a sound are put out of proportion by a level increase.

Noticing that the overall sound grows harsher might just be a byproduct of having grown used to a slightly lower volume while getting EQ settings and whatnot. It's hard for me to listen with objective ears once I've spent a couple of hours getting an optimal mix. (It's always hard to be objective with my own stuff, incidentally -- finding a good pocket for the vocal volume in relation to the rest of the music sometimes takes me hours, because I just can't tell if it's a bit too loud, a bit too soft, etc.)

Just to be complete about the question, I'm not doing anything else in regards to the "Normalize" step -- only raising the volume to its maximum level, so the highest peak in the wave file determines the threshold. I don't apply compression, limiting, etc. during this step. The meter is just a bit in the red during the loudest moments, so it's not out of hand or anything. (Unless I'm mistaken, I don't think digital sound "distorts" like analogue did, unless it's really clipping, which is an entirely different, and quite recognizable, sound.)

Oops. I didn't mean to write a novella. Well, thank you for any help. I should know this stuff, after many years of home recording, but it never occurred to me to ask experts about it, and several attempted searches have yielded nothing.

 

[TalismanRich]

In the digital domain, doing normalization is simply a mathematical exercise. It evaluates the highest peak level, determines how much that peak can be increased to reach 0dB, and then adds that value to all other data points. It should not change the relative values and should not be distorting anything. It's a case of "raising all boats equally" including the noise floor.

However, when processed as something like an MP3, the reconstruction process CAN actually create peaks in "excess" of 0dB, which would obviously be distorted. That's a totally different matter.

What you are hearing is most likely just a difference in perception based on volume unless your audio equipment is distorting. If you decrease the actual acoustic volume to the original level, it should sound the same.

 

[spantini]

When listening to .mp3s on my laptop using BT headphones, the volume seems fairly tame at lower levels. Then as soon as it reaches, like, 76-80, the bass really kicks in and there's a sharp increase in overall volume.

 

[mjbphotos]

What you have to remember is how your monitors, room and your ears affect things, too - the louder the overall volume is, the easier it is to hear to low and high frequencies that may have drop-offs in both your monitors AND your ears.

 

[Chris++]

In the digital domain, doing normalization is simply a mathematical exercise. It evaluates the highest peak level, determines how much that peak can be increased to reach 0dB, and then adds that value to all other data points. It should not change the relative values and should not be distorting anything. It's a case of "raising all boats equally" including the noise floor.

However, when processed as something like an MP3, the reconstruction process CAN actually create peaks in "excess" of 0dB, which would obviously be distorted. That's a totally different matter.

What you are hearing is most likely just a difference in perception based on volume unless your audio equipment is distorting. If you decrease the actual acoustic volume to the original level, it should sound the same.

What you have to remember is how your monitors, room and your ears affect things, too - the louder the overall volume is, the easier it is to hear to low and high frequencies that may have drop-offs in both your monitors AND your ears.

Thanks very much for the help and the input. The above facts are what I suspected, but it would be just like me to overlook some well-known piece of information that others who are experienced in digital recording would already know, like "Duh!" I appreciate your taking the time, fellas.

 

[Papanate]

Will raising the volume tend to change any other properties of a sound?

Think about it - remember the Loudness Controls on stereos? - that boosted your perception of highs and lows to compensate for listening at a lower volume - now are you talking about raising the master Buss or the individual tracks? - The Master will make the whole mix louder and depending on how much you raise things - drive it into digital distortion - the same thing happens on individual tracks - if you push the drums overall the master may distort - or channels will start peaking - you have to put some thought into the raising of volume - how it affects the overall mix or how it affect the master.

 

[bouldersoundguy]

I remember back when Pro Tools used 40 not fixed point processing. Doing things as basic as normalizing did seem to affect the sound. Maybe you need to move from Audacity and Nero to some proper audio production software.

 

[Slouching Raymond]

Once I'm done mixing, equalizing, and otherwise finalizing a song, I tend to maximize the overall volume (by using "Normalize," usually).

So, you're chopping down the volume, with the mixing faders, and then multiplying it back up.
That is throwing away resolution, before trying to re-insert it.
I usually apply normalize to tracks before mixing.

 

[ecc83]

Chris, look up "Equal Loudness Contours" Note that they were derived for steady tones but do have an effect on our perception of musical balance. Your monitoring volume should be consistent day to day, this is why people 'calibrate' their monitors (look for that as well)
You will need a Sound Level Meter but the Keuwlsoft app for phones is fine for the purpose, it needs to be set for the "C" weighting. You will also need a source of Pink noise. Audacity can provide that.

The level you may be told to aim for is around 83dB SPL but for most of us at home this is too loud (83dB is about as loud as a decent FSTV will go before bad distortion) . Yes, calibrate at 83dBC but then back off the monitor pots to a more socially acceptable level (75dB'ish) but mark both setting of the knobs.

MP3 was developed for domestic purposes and tendS to get weird at high sound levels. I am not at all sure about "throwing away resolution"? The R word tends to make the digital audio experts bristle! I am not one so, SHTUMN!

My son often sends me his guitar pieces as MP3 and if I want to modify them at all I convert them to .wav. NO improvement of course but at least I am dealing with 'linear' digital IYSWIM!

Dave.

 

[TalismanRich]

Thin about it - remember the Loudness Controls on stereos? - that boosted your perception of highs and lows to compensate for listening at a lower volume - now are you talking about raising the master Buss or the individual tracks? - The Master will make the whole mix louder and depending on how much you raise things - drive it into digital distortion - the same thing happens on individual tracks - if you push the drums overall the master may distort - or channels will start peaking - you have to put some thought into the raising of volume - how it affects the overall mix or how it affect the master.

Yeah, the loudness button was supposed to correct for Fletcher-Munson effect where you "lose" highs and lows at lower volume. Instead, the common technique was to get things as loud as you wanted with the volume control, then "turn up the bass" by hitting the loudness button. Of course, nobody was trying to listen to natural sounding recordings. Heavy bass and drums is better for dancing and travels through the house better so the whole party gets to listen.

Of course, that is different from normalization, which is just level adjustment, not tone adjustment.

So, you're chopping down the volume, with the mixing faders, and then multiplying it back up.
That is throwing away resolution, before trying to re-insert it.
I usually apply normalize to tracks before mixing.

I typically do the same. I normalize tracks first if needed, then start mixing. It never seemed to affect the sound of the track, but it did help on the few occasions where a particular track was low, and raising the fader didn't provide enough gain to match up with the other tracks. Once I finish mixing, I'll render a couple of times, noting what the max volume level is, then adjust the fader up until I get peaks around -1 or so.

 

[rob aylestone]

I also find that changing volume - as in making it actually louder in the studio makes it sound differently, and I think I've also noticed that my amp and speakers actually sound different at different volumes. I started using control room in cubase for a while basically simply to let me turn up more on quiet sources. what I mean is that if I have a track in the mix, that is right at a very low fader level, on it's own - it was handy to be able to turn the level up so I could hear it properly without pushing the fader. Control room let me add in more volume, rather than go across the room and turn up the amp. Turning up the amp makes it louder, turning up the control room level makes it sound slightly different. I don't know how cubase actually processes the increase, but turning up the amp vs turning up the control room on-screen knob sounds different. Not much, but there. I'm finding that working in 32 bit that actual absolute levels are not vital any more - but the annoyance is waveform displays that sometimes look like empty tracks.

 

[Chris++]

Thank you for the observations and advice, all. I really appreciate that.

(Reading about "Equal Loudness Contours"......interesting stuff!)