Hey peeps,

I produce completely ITB using a PC using cubase 6. I don't do any live recording and mainly produce dance/electronica music. When mixing, I understand that 0db is the limit (in the digital world) I'm also starting to achieve goodish mixes with my master fader reading at around -6db.

I work at 16-bit 44.1, rarely use compression and use HP/LP on every single channel. I'm not great on the maths side of the digital world and so try to keep things simple.

My question regarding gain structure which has been bugging me for a while is that I read threads on the internet that say you should have all your faders at 0db as this is the best resolution and all your signals reading about halfway on the channels. So, after much deliberation about this i employ the method of turning faders down (where relevant) until I get my mixes sounding good and the master fader sitting at 0db and hitting around -6db ish.

SO,is it better to leave all the faders at 0db and then use the gain rotary controls on the top of each channel so that they remain at 0db or very close, OR simply leave the gain controls and use the channel faders to control the volumes?

hope that makes sense? I'm just trying to get nice clear full sounding tunes...

OK, bit of long question sorry peeps!

thanks
Charlie

Quote charlie chalk:

---

I understand that 0db is the limit (in the digital world).

---

Sorry for the pedantry, but it's important when communicating tehcnical issues. You're right in that the maximum digital level available to an A-D or D-A converter is 0dBFS -- but the suffix is important and stands for Full Scale. It's what differentiates 0dBFS from the analogue 0dBu or 0dBV, or even the meter reading 0VU.

Quote:

---

I read threads on the internet that say you should have all your faders at 0db as this is the best resolution and all your signals reading about halfway on the channels.

---

Theoretically, yes. If the faders are anywhere other than at unit (the 0dB mark) then the softwarew will have to start multiplying the audio samples by a gain coefficient. That will produce a longer wordlength which will then need to be dithered back to the final output word length (16 bit in your case). If that process is not handled properly you will incur distortion (ie loose resolution). In practice most DAWs have got this process well sorted now... but it still pays to at least start your mix with your faders at the zero mark since that affords the most precise range of control movements.

The 'signals reading half way up the channels' thing is about maintinaing sensible headroom margins -- like wot we did in the days of analogue. The 0VU mark on an analogue console was typically 20dB below the clipping point -- and that invisible headroom margin is important in maintaining clarity for transient peaks, and providing space for the build up in level when you start mixing channels or tracks together.

Digital meters show the headroom margin that was previously invisible on analogue meters, and tempt us into recording and mixing with far higher levels than we should. Something made even worse by the fact that the headroom is routinely removed in the digital mastering process, so people try to mix to compete in volume with commercial headroom-less reference tracks.

Tracking and mixing become far easier if you maintain analogue style levels, and often plugins and the analogue montoring chain sounds a lot better too. Aim to maintain average signal levels around -20dBFS when tracking, with peaks to -10dBFS or so. When mixing the average and peaks can rise to around -6dBFS -- but avoid going any higher.

When you've completed the nmic and know what the true peak level is you can dial in some gain or normalise the track to close to0dBFS if you wish for better comparability with commercially released music.

Quote:

---

So, after much deliberation about this i employ the method of turning faders down (where relevant) until I get my mixes sounding good and the master fader sitting at 0db and hitting around -6db ish.

---

Sounds a good plan to me.

Quote:

---

is it better to leave all the faders at 0db and then use the gain rotary controls on the top of each channel so that they remain at 0db or very close, OR simply leave the gain controls and use the channel faders to control the volumes?

---

A combination of both. If you track with high peak levels you'll be running the plugins with hot levels too, which might not sound as good as it should. So tracking at more sesible levels, or using level changes sa the first plig-in to bring the levels down a bit is probably a good idea -- and will allow the faders to work around the 0dB position where they are intended to be.

hugh

--------------------
Technical Editor, Sound On Sound

One of the main hurdles when trying to get your head around gain structure if you're new to ITB production is that almost all virtual instrument presets are about 20dB too hot. They do this because louder sounds better, but it's confusing for people when they start out because all the levels they see from those instruments (which they understandably assume to be good levels) are almost (or actually) clipping. Once you back off the output of every virtual instrument so that your peak meters are bouncing about half way down, everything starts to make a bit more sense. Your faders are closer to zero and your mix doesn't clip.

J

--------------------
www.jackruston.com

Hugh,

your amazing! thanks for your reply. very helpful!

I knew I'd get into trouble with the technical talk, sorry! i wish they'd taught more of this stuff to me back at uni when I was studying production...learning the long way now.

thanks again

charlie

Thanks Jack,

once thing I haven't been doing is watching the volumes on vst, most of the time I'm trying to more volume out of em. I will defiantly be watching the levels more as you have described.

My problem is that I have been producing for a while, but no one ever really taught me about gain structure properly and I've been kinda wingin it. after speaking to macc mastering he enlightened me loads and I'm starting to get much better mixdowns. I just couldn't understand how to control volumes without moving the faders.

It's really NOT intuitive completely in the box, and it's a shame you weren't taught this properly. See, if you were recording most of your stuff from a microphone you'd have found it more obvious, because you have a chain of gain structure decisions, the result of which is the final level at digital which you don't usually have direct control over (other than by recalibrating your converter), and you don't need to. You set the mic pre so that you're averaging 0dBVU, or if you choose to run the pre hot for creative reasons you attenuate it afterwards, the signal then hits the AD converter and lands up in the right place according to the commonly employed calibration levels which ensure that the 0VU signal ends up with a similar sort of headroom to that it enjoys in the analogue world. In the computer it's backwards. You pick your digital level arbitrarily, and because a manufacturer can do that, they will almost always choose 'HOT' because louder sounds better. You then come along and try to match those hot levels, and can't work out why things don't sound right. And when you overdub a vocal you're COMPLETELY screwed. So you buy an expensive compressor and U87. But that doesn't work either. So you send your stuff to be mixed or mastered and it's only then that someone tells you what's going wrong.

J

--------------------
www.jackruston.com

Have you got a trim plugin or a channel strip where you can adjust the gain up and down? You need one for mixing ITB really. FREEG by Sonalksis is a good free one.It would normally go on your first insert point. Then as you add more plugins try and keep the signal around the same peak output level
you set.


So if it's -6dbfs and you insert for example a distortion plugin next check the output is not over or under -6dfs, some plugins don't have meters on them so what I often do is temporarily put another instance of FREEG in afterwards just to check the levels.

Hi,

Is there any reason you're not working at 24bit?

I only say this as, following the excellent advice above, if you're working at -20dBFS at 16bit you'll start to creep into the dirty end of noise floors.... If you're going to start using -20 as a 0dBVU reference then really, 24bit would be the way to go...

cheers

Yep, switch to 24 bit (in the Project Set up). I use the gain controls on the channels to bring down the output levels of VSTis then any small adjustments with faders.

--------------------
http://soundcloud.com/richie-royale
http://www.mixcrate.com/richieroyale

I assumed the 16 bit 44.1 comment referred to his chosen destination format. Surely Cubase 6 works with 32-bit float internally?

Hugh

--------------------
Technical Editor, Sound On Sound

Quote narcoman:

---

Is there any reason you're not working at 24bit?

---

What does this mean if he's working ITB with soft synths ?

Presumably the DAW is working at 32float (or something), and "16bits" only applies when he bounces the mix to stereo.

( which I guess might be better done to 24bit if tracks are going to be mastered )

EDIT - hugh beat me to it.

FWIW in Logic (the DAW I'm currently most familiar with), the destination format is normally chosen at bounce down time.

Whereas the "24-bit recording" option in Audio Preferences applies only to recordings you're about to make (so has no effect if using soft synths).

Pretty sure Cubase is similar.

Hey everyone,

thanks for the help n tips n comments

Fell loads more confident now with mixing down as I know exactly where I have been going wrong! Now I need to learn how to get good clean reverbs, I've got reverance and really impressed with it. Anyone got any tips with regards to drums and ambience and how to get that subtleness without loads of reflections?

I work at 16-bit just cos its all i need. I don't bounce to 24-bit unless I know I'm going to send the tune for mastering. plus I'm not really sure how to dither properly and I don't want to screw anything up!

Interesting point about vst presets being way too loud, never thought to worry about this! have downloaded sonalksis freeg as mentioned above, and will get cracking...

thanks again

charlie

Quote chris...:

---

FWIW in Logic (the DAW I'm currently most familiar with), the destination format is normally chosen at bounce down time.

Whereas the "24-bit recording" option in Audio Preferences applies only to recordings you're about to make (so has no effect if using soft synths).

Pretty sure Cubase is similar.

---

First - you'd need to have the FILE format set to 32bit as well. If any audio is rendered at 16bit then you are heading to higher noise floor issues. If the file format is 32bit (and not just the mix engine which is always 32bit - obviously). Okay - lets assume he's keeping every soft synth live...

EVEN THEN - I still wouldn't render a mix at 44.1 16bit in this day and age. Soft synths or not - the noise floor for quantisation still exists.....

SO - the options really are:

32bit file format in the accepted 32bit float engine of cubase
or
24bit audio file format (for such rendered files in the pool or on export) within the accepted 32bit float cubase engine. Especially if we're asking him to mix down at -20 dBFS. On a 16bit export with thats a roughly 13bit mix with peaks into 14 or 15bits. The advantages of 24 bit are even greater when no audio is being recorded - even the best converters are only tracking in reality at an equivalent of 20 bit.... well if we're all soft synth why not take advantage of the super clean and lower noise floor quality!! Know what? Even MORE so in electronic music with the preferences for hyper compression....

Other reasons?
Leaving everything live is risky - render out stuff at some stage.... always a good idea.


Go 24 bit for the mix !!

Reverb in mixing? It's really just a practise game. I fek it up all the time - it's mix 20 that ends up being right!! ..... I really liked Breverb for electronic stuff. And th eLexicon bundles too.... but Breverb is great value.

Quote narcoman:

---

On a 16bit export with thats a roughly 13bit mix with peaks into 14 or 15bits.

---

All perfectly true, and in principle I agree with you, but even 13 bits provides a signal-noise ratio of about 80dB which is more than enough for most domestic applications, and better than most previous-generation replay formats like, er, vinyl records, cassettes, domestic open-reel tapes, FM radio, analogue TV....

I also wanted to comment on the reference to dithering. The internal processing in the DAW will employ 32-bit (floating) wordlengths, so dithering will always be required, whether to generate an output at 24 bits or 16 bits. Any time the wordlength is reduced dithering MUST be employed if you want to avoid low level quantisation ditortions.

In many cases the required dithering is implemented automatically when you tick the output wordlength selection box in the software... but it's worth finding out about for your specific application.

hugh

--------------------
Technical Editor, Sound On Sound