Levels, mixing, and mastering

Okay I had 2 threads running that somehow got all twisted, combined, and taken WAY off course. Let me restate in simple language what the intent was.

Scenario one.

I pick a style, I create a song. I use those 5-6 tracks to make my song work, adding vocals along the way. Then I say "Hey, I think some organ would sound good here." So I go to the next empty track, choose "generate real track", and pick some organ sound. When I generate that track, it is WAY WAY quieter than the ones RB originally created. So my question was "Does the level of the master volume slider, the on the is above the tracks, to the right center, and moves horizontally, have anything to do with the level at which that additional track is created?"

Scenario two.

I have done all of the above, and in most cases for songs I do, I have made gain changes on almost all the real tracks so they almost fill the track but never touch the top or bottom. I want the tracks as hot as they can be without clipping so I can mix DOWN. I would rather have headroom and remove than run out of room to boost.

The original question was not at all about volume, amplitude, level.... it was about how to best use compression and reverb and what "mastering" actually is. I got several great answers that all made sense yet somehow slightly contradicted each other. I find that everyone has "their" special tools that are "the best" (the BEST, Jerry!). And after digesting all of that input, and doing a good deal of reading that was suggested in the thread, I still don't know any better than when I started how to polish the turd.

So maybe I need to ask more specific questions for specific conditions.

My mixes sound nice and bright when I mix down in my studio room. I save to wav, then using Adobe Audition I make them MP3s, one at 320, another at 128 (and if you ever looked at my stuff they all have (hi) or (lo) in the title), take them out to the car and they sound dead, lifeless and drab. No color, all the crispness of the high end is gone.... even when the MP3s sounded fine in the studio. They sound so alive and perky in the studio. In the car, blah.....

So to fix that, do I need to add high end EQ, remove low end EQ, compress, limit, spin, dry, collate, staple, lather-rinse-repeat......? Do I manipulate all the tracks the same? Do the drums get different treatment than the vocals, and the guitar, and the bass? If I EQ the whole thing to find more "snap" on the snare drum, isn't that going to change every other instrument?

I have different EQ and reverb presets for my vocals, drums bass, and instruments other than bass, so I do EQ and apply reverb differently depending on the track content. I don't use compression/limiting at all because I don't know what I am doing and I do more harm than good when I try. That just seemed logical to know that a voice is different from a drum and both are different from a bass guitar, etc.... that thought process may be 180 degrees out of phase. And I do know that from other forum poster's input I can get better, bit I need training if I want to get BETTER. I am just looking for a starting point.

I got the demo of Izotope as someone suggested. Read the help documents for an hour while experimenting. It sounded worse when I finished. FYI, that was in Sonar, as I could not get it to load as a plug-in for RB. Those settings in Izotope made NO sense to me with the cute names they give them. And honestly, since this is a hobby for me, I don't have hours to sit there and play with those presets to figure out what is what. I can't quit my job to play around in the studio writing and producing songs so 25 people can hear them. I'd just like to get a little better so the stuff is presentable. One good thing I DID recently start doing is taking notes (pictures actually) when something sounds right. I guess that's a start.

These threads have really been awesome for the amount of thought and information. I just thought maybe it was time to post fresh and refocus.

I'm always a bit leery of disputing you because you're probably the most respected RB guru around here. Here's my questions:

1. Where is normalize hiding in RB? I just searched Help again and nothing comes up.

2. If RB is not normalizing automatically then how is it possible to mix a bunch of tracks if each one is recorded up at -2db or something? Other software I've tried will simply create an over driven distorted mess and I'm forced to destructively lower each track first. Just moving the console mixer faders won't do it.

3. I asked this in the other thread. If all we have to do is move the faders, then why is it universal among all these articles I've read that digital tracks must be recorded down at -15db or so? What's the big deal with that then?

I just read two more articles that I'm not going to post links to, they still say the same thing. Too hot tracks will overload the output bus and you have to apply either a trim plugin, compression, limiting, something to bring the levels down before you can mix. If you or anybody else has a bunch of tracks that are up there in level and yet you can mix it without clipping simply by moving the faders then one of these things is going on. If you didn't apply a plugin then RB must be doing it automatically someway, somehow.

Bob

Bob.

rharv is right - RB won't normalize anything unless instructed to do so. If you want to normalize a track, or part of a track, you can highlight it and go to EDIT - AUDIO EFFECTS - GAIN CHANGE.

The output of each track is determined by two things - the level of the recording and the channel fader setting. In other words, a hot track with the fader half down will be similar to a quiet track with the fader up full. This principle holds good for both analog and digital.

The trick when mixing is to use the channel faders in conjunction with the meters to ensure that the signal going to the master buss isn't overloading and you can do this regardless of how hot the tracks are. If it's a complex mix we use the sub-groups to make it easier to take sections of the mix up and down. In RB, if the mix is getting too loud you can use the main sub-group (A1) to reduce the overall level.

Having spent over thirty years working with tape, where signal level was our main weapon against noise, I still don't like to see things recorded way down, even though with digital it isn't quite as important. Any decent desk will handle 24 tracks of tape-saturated sound if you watch your gain controls and meters.

There seems to be a lot of dis-information on the internet about levels and mixing, but good principles have always been the same - record as good a signal as possible and keep an eye on the meters when mixing.

ROG.

EDIT - the first studio I helped out in as a boy in the 60s was recording straight into a Ferrograph stereo half-track. The desk had tubes in and a chassis which looked like it had been made out of a railway bridge. The faders more resembled power boat throttles, but it did have a VU meter on every channel. On the control room window was a sign which read - WATCH THE METERS NOT THE PERFORMERS.

Normalisation is provided by the Gain Effect Option 3 which increases the overall gain to -2db. I use this for the final mastered mix in order to achieve a similar level of sound across all the tracks in a CD. Softer sounding tracks can be incresed usually by up to 3 db - more than this I would need to revisit an individual mix.

For recorded tracks within a song, I would not use Normalisation and only use Gain if the track volumn slider is at maximum. As to the concern at over hot mixes which would suggest the need to reduce all track volumn levels, consider using the ALL Volume slider to reduce all the track volumes simultaneously, before increasing low volumn tracks. Don't forget that gain can be added by individual track Compressor/Reverb plugins

For final Mixing I use effects in the Output A1 port - PG-EQ, PG-RTA, Triple Comp, PG-Peak Limit. The Triple Comp plugin, which I have used for the last three years, has proved invaluable. I generally start with the "Preserve Attack" option and study the graphical output to determine what should be done. The Output level determines when to use the ALL slider to obtain the necessary headroom.

I have tried many methods of mixing over the years and, for me, this has proved to be the quickest and best to date.

What I get from both ROG and LynB is there is no global normalization option which is what I was talking about. I'm aware of the option on a track by track basis but what I'm referring to is creating the initial mix in the first place. If you need to mix 4, 8, 20 whatever tracks and overall they are too hot, there's no one global option to normalize them to create a mix. All you can do then is ride the faders. That's what I was asking about.

Now here's where I really don't get it and it's because I completely respect you guys. How can all of these articles be wrong? I try to only read stuff written by someone who looks and sounds like a real pro who knows what they're talking about. Example, the last article I read was linked off of the Pro Tools official forum. It went into great detail about keeping the initial tracking levels low so as to not have the tracks too hot for mixing. Simply riding the faders and saying that is perfectly acceptable flies in the face of this. If that's all you have to work with sure that will work but it's not the optimal correct way to do it if you have control over the initial tracking process. At least that's the way I read it.

Bob

Bob.

You ALWAYS have to ride the faders and watch the meters. It's what mixing is all about and this is why it's a skill which needs to be learnt and practiced. If you have a mix of say thirty plus tracks, it's still possible to overload the master buss even though the tracks may not be recorded hot. If you can't do this, you've already compromised your noise performance, even on digital.

Like I said - there's a lot of dis-information about and some people seem to be trying to make the system idiot-proof. Recording so low that it's impossible to overload is like limiting a car to 25mph so the driver can't break any speed limits.

At the end of the day, I can only say how we've always done it. Other people will disagree and you'll need to come down on one side or the other as you see fit.

ROG

I don't think you are in trouble if you record hot (as low as you don't clip). Once you have all your tracks in there, you then do your gain staging so the master bus is under control. Once that is done you do all your mixing to get the tracks sitting the way you want. Then you apply the compression/limiting on the master bus (or sub-busses) to get the overall song volume where you want it. A little bit of an over-simplification, but there you go.

Here's just the way I was always told to do it.

Record your track this hot

-------------------------------------
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

-------------------------------------

then pull the slider off to 0. Nothing. Dead quiet.

Then start mixing and move the fader accordingly until it blends how you want it.

-------------------------------------

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
-------------------------------------

to start, then up to

-------------------------------------

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

-------------------------------------

and so forth. My thought it that if the starting point (100% of what is available) is low, like

-------------------------------------

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
-------------------------------------

and you start going louder and louder, THAT is where you start introducing noise.

Am I "analog" thinking and "digital" thinking is 180 opposite?

Quote:

applying plugins etc all involve either resampling, dithering, etc and that introduces digital artifacts

Really sorry, Bob, but we're still some way off a meeting of minds here. Re-sampling and dithering occurs only when there is a change in bit rate or frequency and this isn't the case with the trim plugin to which you refer.

The basic trim plugin is just another volume control and has no more effect on the signal than moving the channel fader. It is the digital equivalent of the gain control and pad switch which sit at the top of the channel strip of an analog mixer. These are there because the mixer is receiving signals from an external source, not necessarily the tape recorder and they allow the channel to be calibrated for optimum efficiency.

In a DAW, the mixer is an integral part of the recorder and it's not possible to record anything onto a track which that track's mixer channel is unable to play back. That is, a track which has been recorded and normalized will still not overload the input to that track's mixer channel. The sum of the channel outputs can overload the master buss, which is why watching the meters is so important.

Dithering will occur when a recording is made in 24bit, but the end result is destined for CD at 16bit. This is why some engineers prefer to work in 16bit for CD, particularly as the majority of recorded music doesn't require the extended dynamic range of 24bit. Whether or not the artifacts caused by dithering can be heard is the subject of another debate.

ROG.

you replied to me and quoted someone else there ROG

Popping in here again.

I can't speak for RB's internal audio mix engine, but as far as I know, with most DAW software, when one renders the project to .wav, the output faders are very often automated and by golly they have an effect on how much a particular track contributes to the overall mix level.

I record at 24 bit because it lets me be lazy. I used to sweat the details when all I had available was 16 bit. I would almost always record as close to zero dB on the meters as possible, because on each and every track, that gave me the best signal to noise ratio. With 24 bit A/D, there's an additional 48 dB (6 dB for every bit) of quantization signal to noise ratio available over recording in 16 bit. If I can't make good use of somewhere in that 48 dB then I kind of shouldn't be in the hobby.

The output mix VU meter in any DAW software should tell you if you are into digital output clipping territory. Go red=bad. Stay out of the red=good. It's really that simple. If the render to .wav option inside of RB doesn't simply map this to whatever bit depth and sample rate you choose, then there's a problem.

You should be able to mix to your heart's content inside the PG software. If you have to mix in a different software than what you record, either one of three things is happening: the record software needs some work, or you need to learn how to mix properly inside of it, or there is some feature in your other DAW you should be pounding on PG to include inside RB.

RB, as far as I can tell, is pretty capable to handle the mixing task. Mixing is one of the simplest functions of any DAW software. Why one should have to go outside of it is a bit beyond my comprehension, to be honest.

I think I'm really missing something here. Also, for nearly all intents and purposes with computing power and capability of the last 5 years, you should be able to mix entirely without destructive editing. All of this discussion about the destructive edit aspect of things is one of those things that should quickly fade to the past in your toolbox - that is my experience. It's faded into memory as computing power has increased and DAW software takes advantage of multiple computing cores.

Used to be I did alot of rendering of individual tracks and effects to audio to keep the CPU load down in a project while mixing. However, with my track counts of normally under 12 or so, it's rare that I'm killing the CPU on my lappy which has a quad-core processor, with several VSTi and VST all saying to the CPU "Hey, I need to use you for a second" simultaneously. Keep in mind the effects that are nearly always most CPU hungry are time-based modulation effects like reverb and delay. This is because they require CPU calculation not only on the current sample, but all of those previous samples that are still in the 'tail'. Very CPU hungry. Learn how to bus them to keep CPU happy. Fader changes are just about the least amount of CPU churn that's possible. If I remember my rudimentary digital arithmetic from way back in about 1987 or so, fader changes are simply a gain change which is simply a multiply or divide operation. Bit shifting, I believe. Way simple for the CPU. Guess why multiply and divide were on some of the first digital calculators? It's simple for the calculator. Doing a destructive gain change is just not a good spending of your time and it's destructive. You can't ever change it back - not saying that going back and changing one's mind is always productive - usually it's not.

I read on here lots of folks that say something like: I use RB to record and compose, then I take all my tracks over to XYZ for mixing (insert your 2nd DAW of choice here for XYZ). What a pain in the bum! What is it about program XYZ that is so much better than RB, or PTPA?

I haunt KVRaudio quite a bit and there there are folks using everything from tracker type sequencers to ginormous ProTools rigs and all manner of stuff in between. It's a VERY rare occurrence to read anything there about people regularly transferring projects from one DAW to another just for mixing purposes. Here, it's talked about often. There you'll read more arguments about folks doing tape transfers into ProTools or Logic or something, or mixing using purely outboard busses like the Dangerous audio summing boxes and such, but rarely are there project transfers from one DAW to another simply to mix and render to either 2 track or whatnot.

You must have your reasons for keeping your head wrapped around two different DAW softwares. I have a hard enough time learning one deeply. I'd like to hear what your reasons are because that transfer of a recording project from DAW to a different DAW as at the very heart of this discussion.

For what it's worth, I've collaborated with folks sending me rendered audio from these DAW softwares in the past and used them with zero issue with me providing tracks back to them:

PowerTracks Pro Audio
Mastertracks (I think that's what it was called - it's the other DAW Mac uses also)
ProTools
Cubase
Reason
FL Studio
Tracktion
Garageband
Logic

there may be others, I just don't know. It's never been an issue. Never been a signal to noise ratio issue importing or exporting.

You must have your reasons for doing this. If one is simply just using RB to generate the band tracks, then I can understand the appeal of this specific use, and then using the workflow of a different DAW software. But if you are doing ANY mixing in RB, and then moving to a different DAW, then I don't get it.

Just curious to hear reasoning from anyone who has bothered to read this rather long sermon.

-Scott

Quote:

you replied to me and quoted someone else there ROG

Sorry about that, rharv. It was after midnight at the end of a long day and I forgot to click on the reply box, so it defaulted to you. It was just coincidence that the quote was from another Bob.

I wouldn't try to involve you in a discussion involving dithering - I seem to remember it not being your favorite subject! (Not mine either, actually.)

ROG.

Didn't someone write a song "dithering the day away"?

I'm finding this discussion fascinating. I still wonder then why all the advice from all these pros about recording at such low levels? I completely believe you guys that it's fine to record as hot as is appropriate for a particular part and simply adjust the faders for mixing. I threw out my comment about resampling or whatever just as a guess but apparently that's not it either.

I just did an experiment in RB. Starting with a new blank project I put in one chord on bar one then in track view highlighted the first 12 bars. I then generated a rock Real Drum part and it generated very low, I got one bar at the bottom of the VU meter about -24db. Then I generated a strong RT guitar track, a Brent Mason solo and it came in at -18, much louder on the meter. Now here's where a problem can be, the edit window shows individual peaks at 0 for the drums and -3 for the guitar but that is not reflected in the VU meters. They never went above -18 or -24 and that's where the recording person has to know what they're doing because I saw the exact same thing on my live remote band sessions using my Akai HD recorder. These meters are not sensitive enough. Another example of getting bit by consumer level gear and not pro level gear. I'm sure there's remote equipment with good meters that cost triple what I paid.

These levels that RB generated for guitar and drums are exactly correct per these websites so RB is doing it right. For drums RB is making sure the peaks are not going above zero even though that makes the overall track almost inaudible because the dynamic range on live drums is so high. This is also why a lot of live recordists put a compressor on the drum inputs. I never did, I just applied compression later. RB "knows" that as far as raw tracks are concerned an engineer doesn't want any manipulation going on the he didn't specifically ask for.

In conclusion to all this for Eddie, if you've already laid down some live tracks like your vocals, keyboards and sax and then have RB generate new tracks don't get upset that RB is generating them too low, understand it's actually doing it right. You need to first turn up your monitoring system and then use the mixer faders to pull down your prerecorded tracks, not manipulate the new ones up. At least not yet. A good engineer is going to want to hear what those new tracks sound like in the mix with no changes at first, just raw. Then he'll decide what to do, apply compression, EQ, all of that stuff. And for the future record your live tracks lower to avoid this. In other words follow RB's lead, it turns out Jeff knows what he's doing. And, it looks like I just answered my question at the beginning of this post.

I have to say these discussion with me throwing in some correct stuff and getting others wrong, I've learned a lot.

Bob

Okay so now I get to the climax.....

Let's say for sake of saying that I have recorded drums, bass, guitar, piano, organ, strings, vocals, 2 tracks of BGV, and a guitar solo. As I solo those tracks, they are all this hot on the VU meter.

| |
| |
|X
| |
| |
| |
| |
| |

When I play them back together, will that level stay right there or will the level with all tracks summed become X x 10 and go off the charts?

Bob. I think we're getting there!

Just to expand on your comments about VU meters, all music contains what are called instantaneous peaks which can last for milliseconds, but will effect the overall level. The VU meter is designed to give a more average view of the perceived volume and has a damping factor which ignores these peaks. In effect, anything which lasts for less than about 300 milliseconds won't register on the meter. Some "posh" desks have meters which will switch between VU mode and PPM mode (Peak Program Mode) to enable the engineer to get a more comprehensive view of what's going on.

A small amount of upward limiting can be applied to most tracks to get rid of the instantaneous peaks and increase volume without overly effecting the sound. As you correctly say, RealTracks are recorded to give the engineer the option of treating the track, or not.

ROG.

Eddie.

It's good to have the tracks recorded hot, but yes, if you push all the faders up full you'll have the main VUs off the scale. Just ride the faders and keep an eye on the meters. Alternatively, use the main subgroup fader (A1) to bring down the overall level of the mix before it hits the main output buss.

ROG.

Quote:

I wouldn't try to involve you in a discussion involving dithering - I seem to remember it not being your favorite subject! (Not mine either, actually.)

ROG.

http://izotope.fileburst.com/guides/Dithering_With_Ozone.pdf

Good basic explanation

Ouch, me brain are sore from reading dat! Glad it twere only bay-sick! Me use Dithering but only because it is a default setting in Logic Pro 9. What me know won't hurt I! I go nappy now. Peas Oot!

Page 11 in December issue of Electronic Musician magazine addresses this in a concise answer to a reader question. Assumes 24 bit recording, says record at around -6 dB peak on your DAW's input meters. Also says to mix to -6 dB on the output bus meter to leave a little headroom for changes in mastering.

To address your question Scott about why would someone move a project from RB to another DAW for mixing.

It's workflow for me. Audition does edits much faster than RB does. Example, cutting, pasting or sliding audio tracks. RB does that just fine but does it slowly while Audition does it almost instantly. I could not figure that out until I realized Audition creates some kind of temp files and is working with those initially. When you're finished with your project in Audition and save it that's when you discover it. Audition will ask you what do you want to do with all those edited temp files. You could save each and every one of them into a separate directory if you wanted to. If you don't and just want to save your result as your primary project then you click "no to all" and it will go through a very elaborate "flushing temp files" thing that can take 10-15 minutes depending on how many edits you did. For me working with live band recording tracks it was a ton of edits and one song would take maybe 20 minutes to save. RB apparantly does not do that, it handles the edits in real time so it "appears" to work much slower but then saves the project in a few seconds. In reality when you include the save time in Audition, I think both programs are about the same in the total time it takes to edit, mix and save the same project. It's just that the initial workflow is much, much faster in Audition and I like that.

There's other things like the audio edit window is full screen in Audition, you have much more control over it with the mouse than you do in RB, stuff like that. So, bottom line I stopped importing those projects into PT/RB and started with Audition several years ago.

Bob

Bob, that makes sense. Feature requests submitted?

I think I may have stumbled on to something that has been at the root of a lot of this conversation. We have different schools of thought, but I think there has been a lot of "cross conception" going on here.

I believe we need to differentiate between RECORDING hot and MIXING hot. It seems like different responses are using those two terms interchangeably and that is not what the original question was about.

My past practice (and present practice for that matter) is to RECORD tracks with a good deal of gain on them, so they reach a top level of about -3db. So the gain stage when recording is rather hot, but not reaching 0 or clipping at all.

However, when I MIX those tracks, I pull them down until the channel slider is totally off and then bring them up to where they rarely go beyond -6db or -4db. I record them hotter so there is some meat there if I would want to go any higher when I mix.

Once again I restate this thought and maybe it hasn't been presented in proper context. When mixing down, I can always cut a hotter track, but I can't boost a weaker one. When the slider is as far up as it can go, it's done and then I have to start looking to pull down the other 9, or 12, or 16. Why pull down 9 or 12 or 16 channels because one is weak when, recording that one hotter would prevent that? Digital, analog, digilog or anatal (I made those up!!)... whatever. As long as there is no clip anywhere it is fine.

Thing is, if I mix something down and dump it into a merged stereo wav file, as soon as I get it into Audition I am going to use the gain feature anyway and make it swell up to the full size of the window and see if the VU meters clip or not. I can handle some yellow. Orange is pushing it, but I never go so far where I see any red.

Is that just restating what has been said here or has there actually be a terminology gap about recording hot vs mixing hot?

Quote:

... Once again I restate this thought and maybe it hasn't been presented in proper context. When mixing down, I can always cut a hotter track, but I can't boost a weaker one. When the slider is as far up as it can go, it's done and then I have to start looking to pull down the other 9, or 12, or 16. Why pull down 9 or 12 or 16 channels because one is weak when, ...

I thought we covered this one in the other thread. There are three or four ways to boost a "weaker" track. In most DAWs you can also group tracks together and by moving one fader, ALL the faders will move with it. Can RB do that? It is a quick and easy way to bring down a bunch of tracks at once to start off with a more balanced mix.

Quote:

I think I may have stumbled on to something that has been at the root of a lot of this conversation. We have different schools of thought, but I think there has been a lot of "cross conception" going on here.

I believe we need to differentiate between RECORDING hot and MIXING hot. It seems like different responses are using those two terms interchangeably and that is not what the original question was about.

My past practice (and present practice for that matter) is to RECORD tracks with a good deal of gain on them, so they reach a top level of about -3db. So the gain stage when recording is rather hot, but not reaching 0 or clipping at all.

However, when I MIX those tracks, I pull them down until the channel slider is totally off and then bring them up to where they rarely go beyond -6db or -4db. I record them hotter so there is some meat there if I would want to go any higher when I mix.

Once again I restate this thought and maybe it hasn't been presented in proper context. When mixing down, I can always cut a hotter track, but I can't boost a weaker one. When the slider is as far up as it can go, it's done and then I have to start looking to pull down the other 9, or 12, or 16. Why pull down 9 or 12 or 16 channels because one is weak when, recording that one hotter would prevent that? Digital, analog, digilog or anatal (I made those up!!)... whatever. As long as there is no clip anywhere it is fine.

Thing is, if I mix something down and dump it into a merged stereo wav file, as soon as I get it into Audition I am going to use the gain feature anyway and make it swell up to the full size of the window and see if the VU meters clip or not. I can handle some yellow. Orange is pushing it, but I never go so far where I see any red.

Is that just restating what has been said here or has there actually be a terminology gap about recording hot vs mixing hot?

You are 100% correct in what you're saying here.

Quote:

I thought we covered this one in the other thread. There are three or four ways to boost a "weaker" track. In most DAWs you can also group tracks together and by moving one fader, ALL the faders will move with it. Can RB do that? It is a quick and easy way to bring down a bunch of tracks at once to start off with a more balanced mix.

Yes, RB has subgroup routing ..