RMS - Balanced mixdowns?

Neurojazz

Hey guys,

I thought i might ask about this here, and have noticed some people talking about 'dynamics' and music sounding flat - (this being down to a low signal to noise ratio in an attempt to make a production sound louder/maximised)

I used to get a lot of blown/flat mixes until i discovered the SNR thing, but now wonder how much of a mix should be flattened and what tricks you guys/gals might know for keeping mixes dynamic but with a decent presence.

I get stuck on snares amounst other things where they need delicate compression that doesn't destroy the transients but cannot overtake other items in the mix (i.e heavy dance music which is bassdrum led doesn't need a snare to be more dynamic than the bassdrum)

Most dance music i hear has a SNR of 10db or less (which is pretty crappy and flat) - i'm floating around 16-12db which is making me some sort of black sheep in dance music - which makes the sounds have more impact but alas makes the music sound very different from other dance music productions around it.

Should i go the dark path and over maximise? or is there another way of balancing / getting more out of the SNR? should i have only a few dynamic sounds and flatten the rest?

Usually - for a sound i do this... I shelve 30htz 36dblpf - then i scan and remove mud/odd harmonics - the sound is 'flat' - i then process as needs.... have i killed something by eqing before compression?

I know most this information is genre specific! - but please help!

madtheory

First, if I may be so bold, I'd like to clear up some fog around the terminology.

Signal to noise ratio and dynamic range are similar, but they're not the same. It's desirable to have a high signal to noise ratio, e.g. a good 16 bit system would be 96dB, meaning that the noise floor is 96dB below 0dBFS. 0dBFS is the point where the signal would clip. More bits gets you a lower noise floor, and hence potentially more dynamic range. A big dynamic range is useful when recording live sources (as opposed to synths) because you can use it as headroom to avoid nasty clipping distortion, while staying above the noise floor. That's why the standard peak level for recording in a 24 bit system has now been pretty much agreed to be -18dBFS, which coresponds with 0dBu in a top flight analogue system. Recording at this level in 24 bit makes your final 16 bit mix sound superb. The peak level for a music mix in 16 bit is 0dBFS- and there are many ways to achieve that... briefly-

The whole loudness thing is contentious- there's a war on actually! Generally speaking, a limiter reduces dynamic range without changing the tone, a compressor tends to be more coloured. Compressors also give you more control over how the dynamic range is reduced. Most modern limiters and compressors work by reducing the peak level by a certain amount, so then you can turn the whole lot up by that amount. Older and esoteric devices work by turning up the quiet bits instead.

A high pass filter is not the same as a low shelving filter, although they are very alike. A HPF often has a steeper slope and a fixed frequency. They are useful things to have when there is low frequency content that was not generated by the instrument- eg traffic rumble on the vocal track, etc. LF requires the most energy to reproduce, so it's important to be in control of it if you're being careful about dynamic range. LF noise can eat into that range, noise is bad. So you're on the right track using filters, just be careful where and how you use them.

In the old days, analogue tape gave a natural compression, making recording easier to some degree. With digital, we have to do that manually. If you want loud mixes, there's no quick fix. Sticking a "loudness maximiser" across the master bus is unlikely to get you a commercial sound. You need to decide which elements of the mix need compression or limiting, and which limiter or compressor to use, and how to set the threshold, ratio, attack and release. Every manufacturers' device sounds different, so it's a good idea to have a selection to hand. I'm a big fan of the Bomb Factory 1176 emulation in Pro Tools. George Wohng's W1 limiter is a free VST, it's excellent.

In dance, the pumping effect is created by grouping the drums and sometimes bass together into a compressor. It's likely that each instrument would have its own compression already. The dynamic of the compressor's attack and release curves becomes the musical dynamic. So it's up to you how much it should be. I would be inclined to stick with my gut on this, don't ever try to second guess the "industry standard" or what's "radio friendly". This is art, so make YOUR statement, not what you imagine someone else's is.

Be aware that on club systems and on radio, more dynamic material will sound far better. On radio, it will actually sound louder and clearer than heavily limited material. You might have club DJs complaining that they have to turn your track up slightly more than others, but if your track is filling the floor, I'm sure they'll get over it!

PaulBrewer

Neurojazz,

You're using some terms incorrectly.

Signal to Noise Ratio is a technical term not a 'musical' one which refers to the dynamic range of a 'system' not music within a track as such.

http://en.wikipedia.org/wiki/Signal-to-noise_ratio

Similarly, by 'flattening' do you mean limiting? 'Presence' is a non exact term that came from guitar amps as far as I'm aware, and refers to hi mid frequencies, around the 5k mark (though not exactly!), is this what you mean?

Snares can be heavily compressed without destroying initial transients by varying the Attack on the compressor, a slower attack will allow more of the snares initial transient through, while the compression release will allow you to shape it's 'tail'.

With regard to dynamic range, and while I'm no expert in dance music, I would imagine your tracks need to be as loud as the other tracks around it so there's a strong argument for Mastering in order to match level.

Dynamics is another issue that really is an artistic decision.

There's very little musical info below 30hz. There could be some argument for shelving even higher say 45hz as a lot of PAs will have filters below that.

Unnecessary info in that frequency range (very low) can use up a PAs amp power by trying to amplify something most systems can't reproduce.

Your term 'Odd Harmonics' could be a source of confusion too. I'm guessing in this context you mean 'odd' as in strange? Odd harmonics usually refer to
the relationship of the harmonic to the orginal tone, it's brother being the Even harmonic.

http://wapedia.mobi/en/Harmonic_series_(music)

So..... what's your question again?

Neurojazz

Heya,
So, let me get my head around a few points you made there before i get any more detailed - i'm recording at 48k 32bit float - and mixing down to the same format for later processing into 44k16bit/mp3 etc (but i'll leave that stage alone at the moment)

The -18bd you mention - does that mean i should be mastering at -18db so that renders in 44k 16bit sound better? - i've been told to render at -.3 db... or do you mean that the rms should be around -18db average during louder passages...

In regards to 'pumping' - have no problem doing that

I'll try heavier shelving upto 45hz and see how that affects the mix

Have limters/compressors the whole lot, only missing a few odds and sods.

PaulBrewer

Neurojazz wrote: »

Heya,
So, let me get my head around a few points you made there before i get any more detailed - i'm recording at 48k 32bit float - and mixing down to the same format for later processing into 44k16bit/mp3 etc (but i'll leave that stage alone at the moment)

The -18bd you mention - does that mean i should be mastering at -18db so that renders in 44k 16bit sound better? - i've been told to render at -.3 db... or do you mean that the rms should be around -18db average during louder passages...

In regards to 'pumping' - have no problem doing that

I'll try heavier shelving upto 45hz and see how that affects the mix

Have limters/compressors the whole lot, only missing a few odds and sods.

No, you've got the wrong end of the stick there Squire! I blame Sound on Sound!!

This sounds like one Madtheory may explain better....

PaulBrewer

PaulBrewer wrote: »

No, you've got the wrong end of the stick there Squire! I blame Sound on Sound!!

This sounds like one Madtheory may explain better....

If no one adds I'll drop you a line later..... up to me eyes!!

studiorat

Generally a mastering engineer will ask for 2 or 3 dB headroom on a mix. ie for the mix to peak at around -2dB or so below 0dB on a a digital machine, in order to be able to add eq etc to the master without having to drop the level of the master beforehand.

The -18dB or -14dB thing was what was a standard used to match up PPM's and VU meters being that the ballistics of the two were different. You would match 0dB VU to -14dB or -18dB on a PPM type meter on a Dat machine for instance. It was just a way of setting a reference level.

Bearing in mind you are using synths and samples most of the time, I don't think you should be worrying too much about dynamics etc. Most of the synth lines I come across in dance music have just one dynamic anyway and the loudness of the track would generally equate to the amount of sources in the mix rather than any dynamic control of the performance.

Neurojazz

That's what i'm finding... most the sounds seems very flat apart from 1 or two that stand out and i was worrying that i'm losing dynamics...

I went through some tutorials recently and was shown how to get an average RMS in the louder sections (which i used to get a *blown* sound in - but now more - and for dance music am supposed to aim for 10-12db average - i'm finding that i'm floating around the 16db mark wanting more dynamic elements in the sound and not sure what ones to leave flat etc...

I've found bass i have to leave flat, pads, sfx and the rest are dynamic - basically a more song based electronic format that isn't mainstream dance music - hence overloads, blown mixes and a lot of guess work about what should be limited/compressed and what shouldn't!

My snares used to be flat electronic types but now sounding more like a real livid drummer and the melodies are usually surrounded by counter melodies that take away the dynamics of the main lead so a bit stuffed for a solo sound that is the main dynamic...

PaulBrewer

Neurojazz wrote: »

flat , dynamics...

average RMS, *blown* sound ,10-12db average -
overloads, blown mixes
, main dynamic...

Neurojazz, please define all above Terms! It's very important when talking techy that Technical terms are used correctly.

To extend on Studiorat's and Madtheory's posts.

http://en.wikipedia.org/wiki/Decibel

As a dB is a ratio measurement (0db doesn't exist by itself, only in relation to a defined set of parameters - I seem to remember 0dB VU being .775 Volts across a 600 Ohm load - Open to correction on that one though!)

OdB Full Scale (FS) in digital is an absolute point above which there ain't no more space - There is no +1 dB! That's the the limit, end of story, none louder.

This, depending on gear, can represent + 14 dB, +16dB, +18dB or + 20dB in relation to the Odb (not FULL Scale) point on the meter of a given unit, it's a variable but can usually be adjusted so that all your gear reads the same (Thank Feck!)

Madtheory's point was when you're tracking digitally you leave room for transients so a signal bobbing about the 0dB mark on your mixers VU will represent up to 20dB head room , Comprende?


However in Mastering the rule used be your peak was .03 dB headroom in relation to FS, this was to allow cheaper DAs in home units to playback with out distortion!

Jeez this techy stuff is boring .....

madtheory

There are different kinds of dB. Saying something is at "-18dB" is like saying "that stick is 4 in length". (edit- Paul got in there before me... )

When I said record at -18dBFS, I meant that when you're recording something with an unpredictable dynamic range, such as a vocal performance, then you set your levels while "pretending" that the red is -18dBFS. This allows for any belting that the singer might do in a moment of passion -18dBFS is generally taken to be equivalent to 0VU on a good analogue desk. But if you're staying synthetic, forget about that tip.

There is no advantage whatsoever to recording external synths as 32 bit float, because you can't do better than 24 bit in the real world. But in the virtual world, it does make sense when freezing or rendering virtual instruments. The great thing about 32 bit float is that you don't ever have to worry about peak level. If it's clipping afterwards, you can just turn it down and the clipping goes away. So if you can give your mastering engineer a 32 bit float file, she/ he'll probably be very happy, but check first that they can use that format.

For mastering I prefer if clients give me a 24 bit mix with a peak level between -3dBFS and 0dBFS. If there's clipping, I'll know about it, and they'll have to recall the mix and bring down the master fader. Because most virtual mixers are at least 32 bit float, this works just the same as a 32 bit file.

So, if you want to make your mixes louder, learn how to use compression/ limiting on the elements of the mix, rather than relying on a plugin on the master bus. That doesn'tt preclude bus compression/ limiting/ WHY though! If you're mastering on your own, check out the many tape simulator plugins that are available. Magneto in Cubase is quite good for a few dBs, and the Cranesong plugins are magical. These in addition to the others I recommended earlier. If your stuff is destined for vinyl, then avoid bus limiters altogether, leave that to the cutting engineer.

PS that wiki page has it spelled incorrectly, except for a reference at the end. It's a deciBel because "deci" means "10" and "Bel" is Alexander Graham Bell.

PPS For a final master, 0dBFS is risky as Paul said, because some converters produce an overflow... are crap basically. If you've really squashed the dynamic range, to the point of clipping, then -1dBFS will ensure that the crap converters don't make it sound worse again. But that's for the final master, not the mix.

Neurojazz

In wavelab you can select a loud passage (a fair few bars to get a range) and hit the Y key and this brings up the screen where you can get the average RMS - that was the average rms i was talking about... the principal is supposed to be that if you over compress the average gets lower (so everything is too squashed up against the 0db - there are supposed to be different averages for different genres.... very boring i know, but it's helped clean up my mixes no end so far

I'll probably have another look at the mixdown end and see if i can find the source of what is still swamping the mixes and causing the problem i'm hearing.

..Madtheory... check the link below - i've been using the waves l3-16 to compress/maximise amoungst other kit!

madtheory

Ya, the Waves stuff is good, I find it makes things sound brittle, though that is a popular sound. That little free W1 limiter sounds better, but you'll have to work harder to get the loudness of an L2 or L3. The Sony one sounds better than the Waves.

O ya, why are you recording at 48kHz? That has more disadvantages than advantages, if you're aiming for CD. Stick with 44.1kHz, and you'll avoid a whole load on annoying issues with sample rate conversion.

However, you will find that most of your virtual instruments will sound better at 88/ 96, but that halves your processing power. I always stick with 44.1k, most plugins are designed to sound good at that rate.

PaulBrewer

madtheory wrote: »

O ya, why are you recording at 48kHz? That has more disadvantages than advantages, if you're aiming for CD. Stick with 44.1kHz, and you'll avoid a whole load on annoying issues with sample rate conversion.

Yes, the only caveat being if you're sending to mastering , a lot of houses use analogue gear and have 2 systems that run in tandem . 1 for playback (in your case 48) and 1 for recording the Mastered recording (44.1k 16 bit).

In that scenario 48k is about 10% more info recorded than 44.1khz so that's a good thing!......... Too many bleedin' variables!!

Neurojazz

48k 32float because 1. don't know if songs will end up on CD, but mostly mp3 - and the added possibility that's i'd be playing 48k out live from laptop...

The wave l3-16 has a nice 'velvet' setting - does wonders for me - the 'coldness' i don't seem to have much of a problem with - that may be down to the sample selection and source material i'm picking through

I use the fairlight emulation on the UAD to make things a bit warmer in the final buss with the pultec just before to give the 12k range a nice lift... then it gets taken out into wavelab as a 32bit file for final mastering - that's when the md3 gets pulled out - i have about 6 plugins on the final mix - i might render 2 versions tho (one without compression) for physical pressing as they probably do know a shedload more at the plant about doing that as you said!

madtheory

Paul- Ya, you're right... some places use all digital tho, and some other places use a combination of both. In either case, any mastering house can handle sample rate conversion without any issues. But one doesn't book the place because of the gear, it's the people, right? Right

What I meant to say was, if you're doing your own CD mastering, either get Izoptope or Barbabatch, or just stick to 44.1kHz.

Neurojazz- it's a Fairchild emulation, not a Fairlight. I wish someone would do a Fairlight emulation actually... finally, get the Sony demo and compare it to the Waves... I hear good things about the UAD limiter too.

PaulBrewer

madtheory wrote: »

, it's the people, right? Right

It's the results!!:cool: