Need an explanation of LOUDNESS!!

[ulrichburke]

Dear Everyone.

I hope you had a great Easter - and I hope this question actually makes sense to you - I'm going to over-explain what I'm asking so you can catch my train of thought and tell me if I've got off at the wrong stop.

OK. In a new piece I've got a clarinet - main tune carrier, sounds like a cat with an oboe up its a**e. Anyway. That aside. When I look at its soundwave, the little lines are jumping pretty high. Much higher than those of the backing. Yet it still sounds too quiet. I've tried compression of all types, including blue whale compression (mine own invention, makes elephant's knee look like a blonde in a miniskirt) and you can see how high the soundwaves are jumping, kangaroo on hot coals. Yet it STILL sounds too quiet - and the signals of the combined backing track aren't any louder.

So according to all the science, that clarinet should be audible two blocks away. It should be HUGE compared to the backing (not that I want it huge compared to the backing, I'm just trying to make it sound audible over the backing without EQing the backing to shreds to achieve this.) Despite compression, despite trying turning it up so it's clipping (I know that's wrong, I ran out of ideas) despite listening to it soloed whilst using incredibly scientific methods like twiddling every knob in site, it's not sounding any louder. Its soundwave is jumping like a dancer at a punk rock concert but you'd swear the volume hadn't changed.

Now change the word 'louder' for 'clearer' if you want to. I know EQ is what gives instruments holes to sit in but I didn't want to kill the sound of the backing TOO much. And even when I WAS EQing heck out of the backing to see what would happen - that clarinet wasn't sounding any louder. Except for one annoying bit of it. The transients. They were hitting you like Mike Tyson's lovetaps but immediately after the transient had struck home - too quiet again. Yet the soundwave for the soloed clarinet was still solar hot. EQint the other instruments WAS making it sound comparatively louder - but only by making the other instruments sound like they were being played through a telephone. Judging by the clarinet's signal, you'd've thought it would've been easily audible without needing to do anything to the other sounds.

Now I hit this problem with monotonous regularity in almost everything I write. There's always a couple of sounds, at least, where the soundwaves LOOK a lot louder than the virtual instruments sound. And I've never worked out why. I've even got a couple of magnificent failures somewheres where you can hardly hear anything - yet the soundwaves for the whole file are bright red stratospheric. No, I can't explain that one at all.

But I'm hoping you can.

What is it that controls AUDIBLE sound differences as opposed to TECHNICAL sound differences? In other words, how come you can have something with a red-hot soundwave that still sounds far too quiet - and something else with a much smaller soundwave that's perfectly audible, clear and right in connection with the piece as a whole?

Yours head-scratchingly

Chris.

[Rozovian]

Raise the listening volume, mix it the way it's supposed to sound, then worry about getting the whole track loud.

Not sure what you mean by the waves looking loud; dunno if you're just talking about amplitude or some particular shape or dynamics to them. Check to see that you don't have a lot of excess bass frequencies. Those can cause the level meters to go way up without actually contributing to the sound at all. Other than that, carve with eq, limit transients, compress.

mp3 example plz. Helps a lot to hear and see for ourselves.

[ulrichburke]

Dear Rosovian

Thing is, I'm SOOO noob to mixing I don't really know what I'm talking about either.

For starters, I've put the whole thing up here - the MP3 (Easter Time) and all the individual instrument tracks in a zipfile:-

http://hotfile.com/dl/115852347/1eea20f/Easter_Time_MP3_and_Trax.7z.html

I'm not asking you to do anything with the tracks, just tell me what I SHOULD be doing to them. The idea was to have a backing of nice mixed-together sounds - like a swirl of colours - with a guitar on the left-hand side and a clarinet on the right-hand side taking turns in butchering the melody. It's all VSTs - the guitar and Clarinet are from Proteus VX and the strings and things are all soundfonts. So soundfonts in the middle, Proteus instruments left'n'right. The guitar's fine - might not be the greatest virtual guitar but you can hear it fine. If you look at its oscilloscope reading, it's not that high because it doesn't have to be.

The clarinet's oscilloscope reading is WAY out there. And yet, even if you solo it, it doesn't SOUND that loud. The guitar, with a lower oscilloscope reading, sounds louder than the clarinet!

Which begs the question - if you've got a quiet-ish sound (clarinet, flute, ocarina, things like that) playing the lead melody - how do you keep it audible over the backing? I know that's going to sound stupid to you, because it does to me. It's just - if I get the backing track nice and audible (like a karaoke track, for argument's sake) I can't make the flute/clarinet/similar softer instrument audible over the top no matter what I do to it. And if I use enough EQ to make the lead soft instrument DEFINITELY audible, all the other sounds sound like they're being played through a telephone!

I know it's possible because I've got a TON of flute'n'bad VSTs tracks done by other people. And you can hear the swirly backing with the flute on top perfectly. Nothing sounds like it's coming through telephones on those tracks. I just don't get how they do it!

Yours frustratedly - I WILL get the hang of mixing before I die! The problem with the 'tutorials' is they assume you're doing dance tracks and I don't, I do New Agey tracks so nothing they say seems to apply. Or rather, I don't know which bits of their info to use and which not to.Aagh. My head's a mess with all of this. Just need a bit of help getting started.

Yours respectfully

ulrichburke, wannabe New Age composer!

[Rozovian]

Without getting into individual track mixing, I can instantly see the panning is causing some problems. The left channel is louder. Stuff in center is louder, as it uses both channels, both speakers. You've panned a lot of stuff left without providing enough balance o nthe right side.

Since you provided individual tracks (ugh, slow download), I did a mix of my own. It's not that great, but it's louder and more clear than yours, so I gues that counts as some kind of success. Here's what I did:

The guitar is a lot more clear than your clarinet lead. Time to start carving eq. I was able to boost the track 6dB without the limiter reacting much. Let's go over what I did, track by track.

The Aahs weren't so bad, I just cut 8dB from around 2700Hz to keep them from interfering with the clarinet. The bass wasn't so bad either, but interfering with the cello. I gave the bass a small boost around 100Hz and a small cut around 200Hz. I also dropped it 4dB.

I mono-ed the bodhran as well as limited its loudest peaks slightly. I also mono-ed the clarinet (you're probably working with mono tracks anyway, so just adjust the pan accordingly instead of monoing and re-panning), then panned it slightly to the right (+14, of 64, so slightly). The clarinet then got a boost around 3000Hz by 4dB. It is now too loud during it's higher pitch sections, but it's up to you to figure out by how much and adjust the volume accordingly.

The guitar was too loud, so I dropped it 4dB, and then cut everything above 1kHz by 12dB. That placed the guitar more in the background. The cello got the reverse treatment compared to the bass; with a boost around 250Hz and a cut from 100Hz down, both of these around 4dB. I also cut a little from the 3kHz area here, but that might not make a big difference. Just to make sure the cello doesn't crowd the clarinet.

The Oohs I cut at around 3500Hz by 5dB to make sure they don't interfere with the clarinet. The piano got a subtle boost around 800Hz and a cut around 3kHz. The strings I didn't do anything with. Then I was free to boost the output 6dB in a limiter. Most of the time, the limiter doesn't even have to do anything, so there's some headroom for further boosts if it's still not loud enough.

According to my spectrum analyzer, there's still improvements to be done. The 3-5kHz area could use some beefing up, excess bass could be cut from most tracks that don't need any lows. I didn't bother with reverb (which could place instruments further back, if necessary - like the guitar) and other effects (like a touch of overdrive on the clarinet which would add harmonics, higher frequencies that would help with clarity).

Gotta say that the instruments themselves aren't that great, but we all gotta start somewhere.

[ulrichburke]

Dear Rozovian.

Thank you very much for what you've done for me - I really am touched. As far as the panning goes - I'm using a Soundblaster Audigy LE soundcard. It's only got one hole for speakers. So I had to buy an adapter to be able to plug both speakers in (WHY have only one hole when there's two speakers, Creative Labs!?!) And other people have told me my stuff sounds too far over to the right - when I'm listening to it on my Numark flat sound speakers via the adapter it sounds more central and left. Which is puzzling, but there you go. I tried to compensate but obviously didn't do it enough.

I've got Miroslav Orchestra, but it keeps crashing my software. I've got an M-Audio Audiophile 2496 card but when I had that, its drivers kept giving me non-stop blue screens of death so it had to go and the Audigy was all I could afford. I know it was the M-Audio software that was the problem because as soon as I uninstalled it, no more BSODS. So no more M-Audio, which is a shame because it had TWO speaker holes in it which is more logical.

If I use Miroslav strings, any of them, they stick and stick and stick. I can un-stick them, but the sticking will always get progressively worse until everything crashes on me. It's only the strings that do that but I got so tired of re-setting up eveerything because of Miroslav killing everything that I downloaded the St. John Soundfont Orchestra and used that instead (the other instruments are from Proteus VX). Also, you can E.Q. as you go when you're using individual soundfonts, of course you can't if you're using multiple sounds from the same package, the software only lets you put effects on the package itself, not on the individual sounds coming out of it. (Quick Score Elite Level 2.) Which means you've no real idea what the final piece will sound like as you can't do any ongoing effects. As each soundfont is in its own plaer, it makes life MILES easier. And I've heard sounds as bad/good as mine (depending on how you look at it) on many New Age/Panpipes albums (says he defensively!) You're right, Miroslav would be better if the strings weren't crash-makers.

Did you put the mix you did up anywhere so I can listen to it and hear the difference? I'll try out everything else you say, I'd never thought of any of the things you'd told me.

Yours respectfully

ulrichburke

[Moseph]

re: Miroslav crashing

You might have better luck using ASIO drivers for the soundcard if they're available from Creative Labs. (If they're not, the ASIO4ALL drivers might work.) There should be an options screen somewhere in your program that will tell you what type of drivers you're currently using and will let you switch the drivers if others are available on the machine. Google tells me that QuickScore can use either ASIO or DirectX for the audio. My guess is that you're currently using DirectX, which is way inferior to ASIO.

I'll download the files and take a look at the piece, too.

[mickomoo]

In addition to EQ-ing you can compress or master limit stuff too to give it loudness. But honestly I'm so bad at mixing I really don't know what else to say lol.

[Moseph]

I'm not hearing the problem. The clarinet sounds plenty loud to me. If it still needs to be louder, just pull all the other instruments down.

EDIT: Stop looking at the level meters. Almost the only time the meters matter is when they clip. Other than that just mix with your ears and don't worry about things looking louder or softer than they sound. There are, as Rozovian mentioned, a number of things that influence how the level meters will look -- things such as transient loudness, bass loudness, how EQ is set up, etc. -- and these things may not affect perceived loudness in the same way.

It may also be that the meters in QuickScore are just borked, because when I look at this in Sonar, the clarinet isn't represented on the meter as being louder than the other instruments. Just the opposite, in fact. When all of the level faders are at the zero position (that is, Sonar's playing exactly what you exported from QuickScore), the clarinet peaks below the guitar and the bodhran and in about the same place as the other instruments.

This is what I'm looking at: