Mixing at -17dB is this bad?

[Kozak]

Hey guys,

a while ago I hardly played with the volume faders while producing a track. This ofcourse produced clipping etc..and I started to get more indepth into mixing levels. Using tarekith's site who advises certain volume levels, i.e. -6dB for a kick sound I started to get a finer mix.

I now sometimes notice I start instruments and kicks at even lower volume levels, i.e. -20dB and then the final mix hoovers somewhere at -17dB. I then boost this signal using a utility plugin or a compressor, saturator and maybe finally a limiter.

Now my question is if it is bad to mix lower than -17dB and then boosting the level. I thought maybe sounds or transients or something would get lost....

[muckleberryblim]

An engineer from the UK mastering studio Transistion told me that 'boosting' plugs would introduce a very fine layer of digital distortion. It's not really noticable but it's not good practice.

[Aequitas123]

i've been mixing at -12dB lately and its nice to have so much room. when i'm finished i either bring the tracks up a bit if there's room, or add a very small amount of compression on the Master.

[Kealanm1]

Check out this guide to leveling and mastering it will explain it better than I can . . . its a bit of a read but it will ans your question 100%

http://tarekith.com/assets/Leveling.html

[Kozak]

That was an interesting read! I think I am going to start mixing a bit louder....

[Tarekith]

Try this one too

http://tarekith.com/assets/mixdowns.html

Answers your question a little more directly.

[Fourtothefloor]

In live 8, simply select all the channels and put the fade however many DB.

In live 7 open up the io section in arrangement view by clicking the io button and manually increase the db on every channel.

[fsk]

Interesting, I've only got back to writing music and i have started doing this in the last couple weeks after seeing it on a website also. Do you have any examples of a mix before you adopted this method and a more recent mix?

[Spindrift]

Since Live uses a floating point engine you will not loose resolution if you have the faders on the channels too low and instead boost the master.
So if the signal at the master hovers at -17 dB you can just pull the master fader up.

I guess the mastering engineer mentioned is thinking about fixed bit systems, like Pro Tools.
Using fixed bit you will not only loose resolution, but any operation, even something simple as changing level, will case some rounding errors.

Basically for fixed bit the same rules as for analogue applies...try to get the signal as hot as you can without distorting to get the maximum dynamic range out of your equipment.
IMO it's a good thing to try to stick with the same method in a floating point system, simply because it helps if you one day use another system not using a floating point engine.
But technically it works differently and gain structure is practically irrelevant.

And if you like to stick processing on the master, go ahead, but try to always record a version without as well. Then you or a mastering engineer have maximum flexibility in case you would like to do proper mastering at a later stage.

I prefer to record my masters as 32 bit float. That means that I don't have to pay too much attention to if it's a bit to soft or to loud, and it will still use full resolution and have no clipping. And you will not have to add dithering or truncate the signal to a different bitdepth just for a temporary storage like 24 bit is most of the time.
Then I will do a 16 bit master myself from that file, at which stage I apply compression and limiting. Then the raw master might be sent off to a mastering engineer, and since it's a float file he can set the amount of headroom he want to have for the processing without any degradation of the signal.

[MarkHenryC]

Great bit of info there, Spindrift. Thanks.

[dysanfel]

There are no rules, just guidelines. If your mix sounds great to you clipping the crap out of the master bus, then great. Use your ears.

[michaellpenman]

If you mix at -17dB your stuff wont stand out when it comes yo mastering, you will have to boost the gain so much therefore boost all noise as well.
Also you dont want to clip at all in the digital world

[Tarekith]

-17dB is less than 3bits resolution if you're using 24bit files as the output, so the increase in the noicefloor shouldn't be that bad at all. Unless you recorded extremely noisey signals to begin with.

[michaellpenman]

-17dB is less than 3bits resolution if you're using 24bit files as the output, so the increase in the noicefloor shouldn't be that bad at all. Unless you recorded extremely noisey signals to begin with.

Yeah your correct but i dont see why people need to mix at -17
so your loud elements are at -17 and your quite elements are at -30... seems a bit stupid

[leedsquietman]

Are you talking about peak or RMS levels here ?

If you recorded 24 bit, you have so much headroom that recording very low has no impact, any ME worth his corn can easily pump up the levels without flat topping and without increasing the noise. The extra headroom afforded in 24 or 32 bit float allows this.

A guy on the Cubase forum records his music to an average RMS of -35 to -38 db, with peaks at about -12db to -15db and when it comes out of mastering at around -9dbRMS sounds amazing with no flat topping.

Only 16 bit recording requires you to record hotter levels to prevent the noise floor rising as you increase the gain. Most people these days should be recording 24 bit by default.