reducing or boosting!?

[coi]

I've been getting more into mixing lately as i've been trying to increase my musics sonic power. Noticing that some people state its better to drop frequecies out of a sound if they are not needed and boosting the ones that are needed. Or just boosting the ones that are needed and not lowering other ones.

I am completely aware that this role is quite important in reducing unwanted frequencies in the lower and upper end of the audio spectrum for headroom and more. But is it better to have a sound boosted on the eq or everything else taken out of that frequency and it left the same; (or both)? I've read that boosting is bad and creates new artifacts. I've been doing everything by ear for a while (10 years) but I would like to know how you guys go about this.

Thanks

[daft cunt]

Here is your answer.

[deadly_habit]

save macc some of the trouble
http://www.dogsonacid.com/showthread.ph ... did=585407

[coi]

Here is your answer.

that pdf is f'ing gold mate...thanks

[macc]

@ DH

The only true answer is;

'It depends'.

[slothrop]

Is that the original article that got everyone in DnB trying to "fill the box" with big wads of multiband compression?

[truncated]

i always try to use subtractive eq'ing.

[coi]

i always try to use subtractive eq'ing.

I guess i could have stated that better. But yes subtractive or additive eq'ing, which is your cup of tea folks?

[b-lam]

i always try to use subtractive eq'ing.

I guess i could have stated that better. But yes subtractive or additive eq'ing, which is your cup of tea folks?

subtractive 90% of the time, where it's not possible I do use very shallow boosts, but alot of the time it sounds shit and I end up layering sounds to fill that area...

[james fox]

subtracting above adding, every time.

if you must insist on boosting then use as high quality an eq as possible. also, as a rule of thumb you should boost with a wide Q and cut with a narrow Q...

[jobbanaught]

Here is your answer.

save macc some of the trouble
http://www.dogsonacid.com/showthread.ph ... did=585407

Two nice and in-depth contributions - Thanks mates!

[Sharmaji]

both.

in the digital domain, you can only fill up the sound stage so much that any bit of headroom you gain (before mastering) is good--thus, the more you can turn things down, the more space you can create, the more space there IS to be created IN. get the low-end crap out of your hihats and you have that much more space for your drums and bass to take up.

w/ that said-- sometimes boosting just sounds better...though if yr boosting the fundamental freq's of a sounds, you're essentially just turning up it's volume-- why not just do that, and take out more of what doesn't need to be there?

[jobbanaught]

Nerd mode on:

By mere mathematical theory, boosting in the digital domain can introduce artifacts (i.e. reduce the quality of your sound - "kill the texture"), whereas in the analog domain this can not happen.
Boosting means amplifying certain frequency bands of your signal. In good old analog you have a real signal in terms of a certain voltage, so boosting frequencies is a physical process here. No artifacts, everythings OK.
In digital, the signal is converted (discretized) to an approximation of the real waveform (quality according to frequency and bit depth). Imagine the approximation as a waverform made of lots of tiny little steps, rather than a smooth curve over all frequency bands. Cutting and boosting here is only an interpolation of the effect you would achieve on an analog signal. This is not that much of a problem concerning cutting, since when you reduce the amplitude of certain frequencies, your not likely to be able to here artifacts resulting of the process (because they happen to the parts you cut out). But when boosting, its the opposite, which means the parts of the sound which are affected become even louder! So it all depends on the quality of the interpolation algorithm you have, because in fact when boosting, you add information to the signal. The algorithm needs to smooth out artifacts that might result, because you are adding information without knowing the signal exactly (which is an ill-posed problem). This is the reason why only high-quality digital EQs sound "nice" when boosting, theres a lot of complicated processing which needs to be done to achieve this and not just mangle the sound.

Nerd mode off.

Bottom line: Theres other reasons to prefer cutting as mentioned elsewhere, but technically speaking, if you dont own an anlog or high quality digital EQ, avoid boosting at any cost.

[Hide_One]

Subtractive eq, i find, is easier, i.e. I find it easier to tell where e.qs clash rather than where they need boosting.
But I copy most settings from the cubase studio e.q presets and tweak a little coz theres pretty much everything i need there.

[slothrop]

subtracting above adding, every time.

if you must insist on boosting then use as high quality an eq as possible. also, as a rule of thumb you should boost with a wide Q and cut with a narrow Q...

Isn't that a rule of thumb from recording bands where you want the instruments and vocals to sound as natural and lifelike as possible - IIRC it works out that a wide cut will chop out a lot of the specific character of the sound and make it sound distinctly different, whereas a narrow boost will cause a particular frequency to stick out, which will again sound 'unnatural' on a guitar or a voice. But when you're talking about wobble bass and synth claps, worrying about 'natural' sounds with their original timbre is kind of a secondary concern. Albeit still worth thinking about if you want to keep the sounds sounding the same.

AIUI if you're working in 24 bit there's no practical difference between boosting the high end by 4dB and then reducing the overall volume by 4dB and cutting the low end by 4dB, assuming your curves match up. I tend to do whatever is the more natural description of what I'm doing ie if I want to add a bit of high end sizzle to my drums I'll boost a bit, if I want to take out the bit of kick that clashes with the sub I'll cut it.

But I might be talking out of my arse here. If someone like Bob or Jason wants to put me right I'd be glad to know...

[Sharmaji]

AIUI if you're working in 24 bit there's no practical difference between boosting the high end by 4dB and then reducing the overall volume by 4dB and cutting the low end by 4dB, assuming your curves match up.

different plugs (and different hardware) sound different, obviously; do a hard, 48db/octave cut somewhere and some plugs will build up resonance around that area... some won't. w/ some kit, a 2db boost will sound like a 12db boost on others. all depends on what gear or plugs you're using.

case in point-- i used to hate the cubase stock eq because it made things sound really weak; any time you boosted or cut something, you wound up messing up the phase relationships at key points. the waves Req made a huge difference by keeping things a lot more kosher.

[macc]

case in point-- i used to hate the cubase stock eq because it made things sound really weak; any time you boosted or cut something, you wound up messing up the phase relationships at key points. the waves Req made a huge difference by keeping things a lot more kosher.

I hope you don't mind if I pick up on something there. Before I start I want to say I agree that I strongly dislike that eq too! However, it isn't 'destroying phase relationships' or any such hocus pocus.

I suppose the simplest way of putting it is that there isn’t one law of physics for one eq and one for another. There is only frequency, amplitude and time/phase. Ignoring ‘analogue’ emulation (ie inherent curves with all controls flat, distortion emulation etc), all digital minimum-phase eqs will sound the same for a given curve. If you match the applied amplitude vs frequency curve in two eqs, the phase response is the same. It’s not one eq or another, it’s physics If you work hard enough you should be able to get things to null to a very very low level, if not completely. Assuming you have the time to do that of course.

Naturally, most eqs have different designs/applied curves and as a result, seem to have a different inherent ‘sound’. But one isn’t doing anything magic or nasty or ‘destroying phase relationships’ –just applying different curves. I *imagine* - but would need someone who knows about dsp to confirm this – that certain ‘nice smooth analogue blah blah’ curves are more processing intensive to achieve than others, which explains some of the differences. This gets into filter design and implementation though and I’m not an expert there to be honest!

Again, linear-phase eqs and analogue emulation etc etc are besides the point here, we’re talking normal standard digital equalisation. There’s also one or two other that frankly I can’t be bothered to get into, but they don’t affect the essence…

Point is – choose one you like the ‘sound’ and features of, use it, don’t worry about it.

AIUI if you're working in 24 bit there's no practical difference between boosting the high end by 4dB and then reducing the overall volume by 4dB and cutting the low end by 4dB, assuming your curves match up. I tend to do whatever is the more natural description of what I'm doing ie if I want to add a bit of high end sizzle to my drums I'll boost a bit, if I want to take out the bit of kick that clashes with the sub I'll cut it.

AFAIK you’re right, apart from that the overall/fullband level correction would not be 4dB. The point stands though - as discussed above if your curves match *exactly* and there’s no ‘analogue’ blah blah happening, you have no right to expect anything other than the same result. It’s not *really* anything to do with 24-bit, though I see why you mention it

Sorry for the long waffley post.
What was the question again?

[slothrop]

AFAIK you’re right, apart from that the overall/fullband level correction would not be 4dB. The point stands though - as discussed above if your curves match *exactly* and there’s no ‘analogue’ blah blah happening, you have no right to expect anything other than the same result. It’s not *really* anything to do with 24-bit, though I see why you mention it

Yeah - aiui there's no mathematical basis for the "boosting introduces artifacts" thing. Or at least, there is, it's just that everyone tends to visualize digital audio as being approximately 4 bit fixed point, whereas at any bit rate you're actually going to work at and in floating point, afaict, the 'artifacts' are not so much "insignificant unless you're a perfectionist" as "basically infinitesimal." Moreso if you're cutting everything but the bit you were going to boost and then turning the whole thing up because someone on the internet told you to"always cut and never boost."

I'd have to think it through properly, but basically the only argument I've heard for boosting causing digital artifacts would also apply to 'prove' that turning something up causes digital artifacts.

[macc]

^ This is all true.

The reasons are all more musical than anything.

I'm going to stop because my G14 is currently making me rather moist, and I am liable to go on about it.

[jobbanaught]

Yeah - aiui there's no mathematical basis for the "boosting introduces artifacts" thing.

It maybe just depends on your point of view. What I wanted to point out is that boosting just certain(!) parts of the frequency band can introduce artifacts (because interpolation is applied at the boosted band). Thats a whole different story than boosting all frequencies. Also my post was more from the theoretical point of view, passing on what I learnt about that matter in lectures on signal processing.
In practice it occured to me that boosting using "cheap" EQ (like the one in Live or maybe also in Cubase) justs sounds shitty. So I thought it was for the above reason. But maybe I just thought it sounds shitty because I knew that by theory it *could*. So it would be more of a self-fulfilling prophecy...