Perception of loudness, stereo balance: peaks vs. averages/RMS

[eq1]

I find myself often struggling to determine the correct levels (pan and/or gain) for left and right channels, so that the stereo balance is equal.

For example, at the moment I'm listening to a stereo piano track. Panned to center, RMS meter shows equal values (about -33.4db), peak meter shows left channel about 1db lower than right. It doesn't sound balanced to me.

I need to pan it about 8% right to achieve the perception of balance... When I do that, though, the RMS meter shows the right channel about +1db vs. left; on the other hand, the peak meter shows left and right channels at same value...

It looks like my perception of 'balance' has more to do with the peaks than the average loudness, at least in this case.

I've tested my ears a few times, they are slightly different, where my left seems to be more sensitive in the 2k-4k range, whereas my right is more sensitive at 500 and 8k... Not sure how much that factors-in. I don't think my headphones are inaccurate, and though my headphone amp can be a bit off, I don't think that's the case here...

So, what? Is there some auditory perceptual mumbo-jumbo concept dealing with peak loudness and average loudness and their impact on perceptions of stereo balance, or loudness in general?

Is this normal - that I might perceive 'balance' based on the peaks rather than the average level? Or maybe my ears and/or brain are screwy?

I think this latter question is what I'm really trying to figure out. I'd assume I just go with my ears, what sounds right, and forget the meters. But, I don't want to do that if it's very likely that I have a hearing or brain problem... If y'all say that that's normal - that it's normal for peaks to influence perceptions of loudness/balance more than average levels - then I can move forward in life. If not?...

[karbomusic]

I don't know about the peaks and all that but when I calibrate my system, one of the steps is to use an SPL meter with a test tone to ensure the exact same SPL level from each monitor. If that is spot on, ears should agree if there are zero hearing issues.

If my hearing (or lack of) doesn't agree then I'd just lean my head (or whatever) when it matters so that I'm compensating for my own perception without affecting others.

I suppose one could use Monitoring FX to achieve similar with headphones if the engineer's hearing is favoring left or right? That said, that's about as far as I've ever dug into it out of fear of getting myself into the weeds of diminishing returns.

[Philbo King]

Loudness is primarily indicated by RMS level, and peak levels are of little use in determining loudness. It is quite possible for the left and right channels of a stereo track to have very different crest factors. A very high crest factor means the peak level could be quite high while the perceived volume is very low.

In general it's a good practice to calibrate your playback system to give equal left & right sound pressure levels (as measured by a SPL meter) at the listening position using pink noise. Then mix with your ears, not your meters. No meter can tell the whole story...

[ashcat_lt]

Well there are a lot of things that can go into perception of loudness. RMS is not the full answer. LUFS (basically gated and filtered RMS) might come closer, but there’s a lot more to it. Specific to the OP, a sound that is slightly more dynamic - especially if it has a sharper transient and/or a little more high frequency information in general, will tend to sound closer, which is almost assumed to be louder. Or more like it is preferred. It draws more attention so feels louder.

Yes properly calibrated monitors in a decently treated room are always important.

[eq1]

I haven't gotten around to buying an SPL meter and calibrating my monitors - something I'm now more eager to do (I actually just recently took care of a defective monitor, that was the first order of business). In this case though I am listening on headphones...

Any advice, pointers on what SPL meter would be adequate? I perused amazon and saw a price range of about $20 to about $80. I searched online and read a few things, gathered that I should get one that has C weighting, possibly with a frequency range at least up to 8k? I don't use a smartphone so apps are out of the question. Do I need to spend more than $20 to get one that will likely be good enough, for instance, or will it likely depend on the specs? Etc...


From what you all say, it sounds like there can be multiple factors that impact the perception of stereo balance. That's what I was thinking at the start, peaks being one factor. It sounds like, it may be the case that one side is technically louder than the other, when measured in, say, LUFS. BUT, one doesn't necessarily strive for equal LUFS on left and right channels to achieve 'balance'. Does that sound right - a fairly solid, general summary of the idea??

IF you are indeed trying to have balanced loudness left and right you would 'look at the meter'; if you're trying to achieve the perception of a balanced stereo mix, you have to have calibrated monitoring, a good listening environment, and use your own ears and judgment. Good summary? Yay/nay?

I guess I'm trying to get a better understanding of the relationship between 'loudness' per se, as measured in say LUFS or RMS, versus the perception of 'balance'. I'm leaning toward the ideas summarized above, where basically you can have a lot of things that impact the perception of balance, average loudness being one of them, peaks being another, probably high vs. low frequencies being another, just lots of things...

[karbomusic]

C weighting and slow usually. Your choices are a more expensive one with an official calibration, or a cheaper one that is "close". Since most of what you are measuring is the difference between monitors etc., the SPL number being exact isn't so important as the difference between measurements - and since you aren't certifying anything a dB or so off isn't going to be a big deal.

That said, get one, it's a lifetime purchase and will serve you for a very long time. It's also nice to use one enough that your ears start learning what dB SPL 'something' actually sounds like.

[ashcat_lt]

I found an app for my phone. The free version doesn't do C weighting, and I've never bothered trying to calibrate it, but it's certainly close enough for the rough balancing. I happen to have an old Radio Shack standalone unit that I use when I want to actually set my listening levels Katz style.

Honestly, though, if we're talking about relative levels rather than absolute SPL, you don't need an SPL meter at all. Quick and easy, just grab your nearest mic. Flatter and less directional is better, but whatever will work as long as you put it exact dead center where your head should go. Send white noise through, switching from speaker to speaker, and record the result in Reaper. Split the resulting item where you switched the panning. Use your tools to analyze the results. SWS Loudness tools, Spectrum Analyzers (ReaEQ if nothing else), etc. Adjust...things...til you're satisfied that they match as well as you'd like based on whatever metric(s) you find important.

Best would probably be a measurement microphone (they don't have to be expensive, and can be useful for other things) and use something like REW to shoot the room and actually sort it out. Uneven reflections can really mess you up, and even reflections if they're strong enough and fast enough can smear your perception of the stereo field altogether.

[karbomusic]

Originally Posted by ashcat_lt

Honestly, though, if we're talking about relative levels rather than absolute SPL, you don't need an SPL meter at all. Quick and easy, just grab your nearest mic. Flatter and less directional is better, but whatever will work as long as you put it exact dead center where your head should go. Send white noise through, switching from speaker to speaker, and record the result in Reaper. Split the resulting item where you switched the panning. Use your tools to analyze the results. SWS Loudness tools, Spectrum Analyzers (ReaEQ if nothing else), etc. Adjust...things...til you're satisfied that they match as well as you'd like based on whatever metric(s) you find important.

That would work fine but seems like a lot of steps compared to turning on the SPL meter and looking at it, so next best thing I guess.

I would imagine the OP would eventually want to use REW and start shooting their room for all the reasons the rest of us do. Even then though, it's good if they have an SPL meter to calibrate REW to so it's readings actually reflect real'ish SPL levels YMMV.

[DVDdoug]

I've tested my ears a few times, they are slightly different, where my left seems to be more sensitive in the 2k-4k range, whereas my right is more sensitive at 500 and 8k... Not sure how much that factors-in. I don't think my headphones are inaccurate, and though my headphone amp can be a bit off, I don't think that's the case here...

In this case though I am listening on headphones...

With speakers your brain "knows" that your ears don't match so your brain will probably automatically compensate and know where the center is. Possibly with headphones too if the difference isn't too extreme.

Check your ears & monitors (and headphones) with a mono file which should play identically through both sides and hopefully it will sound balanced/centered.

Normally it's done by ear but "technically" it's super-complicated because in a normal stereo mix the left might be louder one moment and the right louder the next moment and they don't have to mathematically balance-out or average-out. It just has to "sound right" and the main vocals should sound centered.

If it turns-out that you can't trust your ears you might need to figure-out a balance control to compensate. If your monitors have separate volume controls, that's easy.

Or, maybe match the RMS or LUFS levels instead of relying on your ears.

Most commercial releases are limited so the peaks will match (usually normalized to 0dB) but I wouldn't expect the RMS or LUFS levels to exactly match.

...Your SPL meter doesn't have to be accurate. It just has to give the same reading if the volume doesn't change. And just moving the SPL meter slightly can change the readings because of room reflections so you probably won't get "perfect" results anyway. At one time I was doing some experiments in my (untreated) home office with high frequency tones and when I moved around behind the SPL meter (on a mic stand) the readings would change by a few (or several) dB.

[ashcat_lt]

Originally Posted by karbomusic

That would work fine but seems like a lot of steps compared to turning on the SPL meter and looking at it, so next best thing I guess.

It doesn't actually have to be more complicated than recording a sweep from hard L to hard R and then looking at the master meter, honestly. My thing was these are things most of us already have, and allow a lot more in depth analysis in case you want to go there. If one knew all of the specs of the microphone and interface, one could even calculate the absolute SPL without too terrible much trouble, such that a dedicated meter isn't really needed at all.

Basically, you don't have to go buy anything to get a good quick idea of where things are at, and if you need more than that, you'll want more than an SPL meter anyway.

[karbomusic]

Originally Posted by ashcat_lt

It doesn't actually have to be more complicated than recording a sweep from hard L to hard R and then looking at the master meter, honestly. My thing was these are things most of us already have, and allow a lot more in depth analysis in case you want to go there. If one knew all of the specs of the microphone and interface, one could even calculate the absolute SPL without too terrible much trouble, such that a dedicated meter isn't really needed at all.

Basically, you don't have to go buy anything to get a good quick idea of where things are at, and if you need more than that, you'll want more than an SPL meter anyway.

I didn't say complicated... it was just a lot of steps vs pushing ON on a physical meter and looking at the output - it's a handy tool just like your laser distance measuring device - not a big deal really.

That said for most people 20/30 bucks for a meter they can use for life and anywhere/everywhere isn't a must have or a regretted purchase. Fuck, I'll buy it and ship it to him.

[ashcat_lt]

Yeah I suppose, and little known fact but some SPL meters actually are perfectly usable measurement microphones. My RadioShack unit has a dedicated output for use in more exacting measurements. But again, I already own that. Searching for, ordering, waiting for delivery...that's quite a few steps.

[eq1]

Thanks for all the feedback, I appreciate it. BUT...

Originally Posted by eq1

I haven't gotten around to buying an SPL meter and calibrating my monitors - something I'm now more eager to do (I actually just recently took care of a defective monitor, that was the first order of business).

How come no one told me to knock on wood? I flipped ON my monitors and my left one is flaking-out again (tweeter keeps dropping out). It's been weeks since I 'fixed' it, with no issues. Guess I'll have to open it up again...

[karbomusic]

Originally Posted by ashcat_lt

Yeah I suppose, and little known fact but some SPL meters actually are perfectly usable measurement microphones. My RadioShack unit has a dedicated output for use in more exacting measurements. But again, I already own that. Searching for, ordering, waiting for delivery...that's quite a few steps.

I have the RadioShack one still, got it in like 1990 or something. Yea, it has the output to use it like a mic and IIRC REW has a setup for using it that way, and I have a generic cal file for that model.

And I have an off-brand SPL meter I got a couple years ago but for REW stuff I now use my SonarWorks mic(s) other than using the SPL meter to calibrate SPL in REW so that the displayed SPL in it's measurements are halfway accurate - I also have a couple of ECM8000s but some info on the net and measurements I just made last night, lend to the idea that it struggles for accuracy above 1kHz.

The Sonar Works mics, which were less expensive than the ECM8000s have a per mic calibration profile which seems to be significant thus far. One of my Sonar Works mics is defective, they sent me a replacement but might be able to fix the bad one, it's problem is a 60hz hum all the time. I haven't pulled it apart yet though.