What Recording and Mixing Levels preferred?

I've heard various things about optimum recording levels for Sonar, but would like to know what others aim for. One person in this forum shoots for peaks at close to 0 to track, then mixes down to -10.

My guitar teacher, who thinks he's the ****, says to track at -12 for headroom. Whatdya think? Anybody use limiters to set their tracking peaks?

Bill

Generally you should record as hot a signal as possible without distorting it.
My guess is that you would get a very low volume recording and mixing down
to a max level around -10dB.
That's why you use compressors, to lift the overall volume while keeping tops from peaking over 0dB.
3-5 db is enough headroom I believe.......

I generally aim for the -5 to -10db range when tracking, although sometimes that's not enough, particularly with inexperienced performers who don't know how to give you a solid level when sound checking.

As for mixdown, I just try to keep it under -0.1db. Sometimes if I don't have the time to play the song one more time before exporting to WAV for mastering, I'll cheat a little and place a limiter like Waves L2 on the main out bus and have it do nothing but ensure -0.1 is never exceeded. i.e. - no gain, only reduction. Still, when using this shortcut technique, you don't want to be sending overs too often anyway. The idea is not to apply limiting, but merely to ensure distortion-free output. Generally turning down your master bus level is a good idea if you're close.

As for using limiters during tracking, I have a drawmer mx30 outboard comp/lim, but I don't use it for limiting. I use it for compression only (I find the limiting has too many artifacts and sounds unnatural). In addition to that, I will sometimes set my pre-amp to do some tube compression on the input stage (assuming a little crunch is okay for that signal). But no limiting for me during tracking.

Similar thoughts about this here. I track as hot as possible, and when I am ready to mix I paste -10 into all the TRACK levels and start there. That way I don't end up fighting with the top limit of the faders while mixing. I mix to several busses in groups (drums, vox, guitars, and so on) apply some light compression and limiting (and EQ and verb or whatever), and send all the mixbusses to a single output and aim for that to be around -5 to -.5. If I am going to bounce digitally I make sure I dont clip and bounce. I do any 'mastering' (quotes since usually I would prefer tracks to be mastered elsewhere, or will save the heavy work for when the track is done). Sometimes I will send the mixbusses to groups on my mixer and sum them on the board sending them back to a channel that records the output.

When I wasn't quite as good at mixing and had the sort of problem where you turn up all the tracks over and over until you have no more headroom I would even actually group all the faders (relative mode) and pull them downagain retaining my mix and gaining headroom. That was enough trouble to help me learn not to get myself in that position. ;)

cAPS

ORIGINAL: Saxon1066

I've heard various things about optimum recording levels for Sonar, but would like to know what others aim for. One person in this forum shoots for peaks at close to 0 to track, then mixes down to -10.

My guitar teacher, who thinks he's the ****, says to track at -12 for headroom. Whatdya think? Anybody use limiters to set their tracking peaks?

Bill

First of all gutiar teachers always think they are the sh!t.. LOL. You don't have to use a number for this. As long as you don't clip (go over 0) you will have a usable track. Now, if you are recording at 16bit you will want to try to keep the signal as hot as you can safely becuase the quiet end of 16bit can get a bit grainy. At 24bit you can record at just about any level and still get good quality sound. It just might make it hard to mix if you get one that is too low, so again just keep a good level and you will be fine. As for mixing it depends on what you are going to do for mastering. If you are going to have someone else do it you might want to leave them 10db or so. But keep in mind the sound and even mix relationships can change when you apply a master compressor. If you have a DAW like Sonar it's better to go ahead and put the master compressor on the stereo buss while you are mixing so you hear what you are going to get. Mix as loud as you want the final product to be without cliping. If you miss by 1 or 2 db that's OK, you can normalize that up later but no more then that or tone and mix will change a little. With a DAW it's no be deal to re-run a mix so go for it.

If you record at 16 bit it's worth taking the time to really get the level as high as you can before clipping. With 24 bit, you get so much extra headroom that tracking with peaks at -6, -12 or even -20 is fine, but it will make mixing easier if all of your tracks are at similar levels. for a variety of technical reasons, recording hotter (at 24 bits) isn't going to result in appreciably better results if you're going to be mixing tracks and mastering to 16 bit (for CD). So if the source is really hard to control, like clean guitars or amatuer singers, go ahead and leave enough headroom to be safe. Cheers.

"... it will make mixing easier if all of your tracks are at similar levels..."

This is why SONAR has a Gain slider, in addition to the standard Volume slider for each track. If you have level differences between tracks, use the Gain slider to bring each of them to nominal levels before mixing. The Gain is your "pre-effects" volume, so if you're not using it before mixing, chances are your effects aren't getting the proper level of signal to maximally hear their results. After your Gain is adjusted, then your effects bin, THEN use the Volume slider for mixing and all's well.

Waltenstein: And the Gain slider has nothing to do with recording input levels, right? (Dumb noob question.)

Ohhey: I'm wondering why you need to mixdown to -10db for premastering. I'm planning on sending my mix out to a mastering house. What do they do that needs so much headroom?

I'm definitely using 24 bit. Why not? My problem is a singer and a guitarist who perform at wildly varying levels. They're good musicians, very expressive, but the quieter passages get lost, and they "can't help it." So, if I set the peak levels even close to 0, there are still too-quiet phrases.

Maybe a compressor on the signal going in?

I tend to approach each track with a level that will fit in the mix better down the line. In other words, I try not to make each track so hot that I later have to reduce the level anyway. We're working in the digital domain, so as long as you get a 'good' signal (meaning not TOO low, not TOO high, avoid clipping for sure), levels are managable in many ways.

I tend to mix as I go, so to speak. IOW, I set levels based on the sound I'm looking to end up with. I usually have a 6 bus drum setup (4 busses for various parts, like Kick on one bus, cyms on another, toms another, etc) and they all go to a 'drum bus' (composite of the other busses) plus a drum aux bus (sometimes more than one) which is used as a send (insert) for FX, again going to the "drum bus" composite - which then goes to a Master Bus (as do my other busses or outputs). The Master goes to the hardware audio.

With that setup - I tend to keep the drum mix in the -6 to -10 range. I usually stay around -6 for other tracks as well, so that the "master bus" doesn't get saturated. I also almost always start with a master limiter of some type (classic master limiter, PSP Vintaage Warmer, etc) so that I don't blow anything out while recording or audition plugin sounds.

But, everyone has to develop an approach that works for both the particular project and an approach that works for your particular workflow style, IMHO.

ORIGINAL: Saxon1066

I've heard various things about optimum recording levels for Sonar, but would like to know what others aim for. One person in this forum shoots for peaks at close to 0 to track, then mixes down to -10.

My guitar teacher, who thinks he's the ****, says to track at -12 for headroom. Whatdya think? Anybody use limiters to set their tracking peaks?

Bill

The Gain slider does not affect the recording level. When you record a track you're creating a WAV file. This WAV's contents is dictated by your signal path, including your mic, preamp, hardware insert effects, and your A/D conversion. If you want your WAV to be hotter, you need to turn your gain up on your hardware preamp, reposition the mic, turn the instrument up, etc.

When you're mixing, you're playing back all your already-recorded WAV files and balancing them against one another. The Gain slider adjusts the volume of the WAV that is fed into your effects bin for that track. Then your effects do their thing and (hopefully) don't change the nominal volume level by much. Then you use the track's Volume slider to balance your track against all the other tracks.

ORIGINAL: Saxon1066
I'm definitely using 24 bit. Why not? My problem is a singer and a guitarist who perform at wildly varying levels. They're good musicians, very expressive, but the quieter passages get lost, and they "can't help it." So, if I set the peak levels even close to 0, there are still too-quiet phrases.

Maybe a compressor on the signal going in?

Compression is not the best tool for correcting this. This is exactly what you want to use the gain control for. You can either draw in or automate a gain envelope to change the relative levels of different sections of the song, then process, edit, and mix to taste.

You could also try tracking the song in sections, with different levels, possibly even different mics for quiet vs. loud sections.

Some prefer to use light hardware compression or limiting on the incoming signal to even things out before it gets to disk, but that works better for smoothing over incidental inconsistencies, not for clobbering whole passages of songs with wild level changes.

You can use compression to to great effect to change the dynamic "feel" and "pulse" of the performance, but it doesn't usually give very good results when used as a sort of automatic volume control. Cheers.

I have a question relating to all of this... From what I can understand the RECORDING LEVEL is not based on using any sliders in Sonar 3, but rather adjusting the levels on the input device (mine being a Motu 828 MK2). That said, when doing the initial recording should I leave the faders for each track at 0db? And then just keep an eye on the graphical display of volume levels, and THEN if I see it creeping up over 0db I decrease the gain in my Motu 828 MK2?
Because If I lower the fader down to -10db or -5db in Sonar, does that REALLY effect the actual volume of the recorded track. I mean, if I slide the fader up wouldn't it just start to clip if the signal was too hot from my Motu?

I'm full of dumb questions this month... heh
Dave (RRabbi)

Your settings in the SONAR track will have zero impact on the recorded sound. The only thing that matters there is your input hardware, and the settings on your sound card if they add any gain/reduction.

That said, when you record in Sonar, do you leave the faders at 0db, and then ride the gain on the input device (in my case the Motu 828 MK2) to make sure it doesn't clip? Or is there an easier way to do this?

Dave

Ohhey: I'm wondering why you need to mixdown to -10db for premastering. I'm planning on sending my mix out to a mastering house. What do they do that needs so much headroom?

They don't need headroom: The key is not to overcompress your mix prior to mastering. Preserve dynamic range - which means don't turn down the volume unless you have to to avoid clipping.

Any mastering house can turn down the digital volume later, but intentionally reducing overall volume is intentionally losing dynamic range. Of course, that said - you have a buttload of extra dynamic range when working in 24bit and targeting 16 bit, so in most cases it won't matter. if you are using 16 bit, though, squeeze out everything up to -.1db

Eric

Here's something I always wanted to know: on commercial rock recordings, I've noticed (I think) that when the vocals come in, the levels of all instruments drop back into the background. Ten, during instrumental passages, like between verses, the instruments come back to the forefront.

How is this achieved? Do the mixers use automation (or in the old days, a steady hand) to drop the instrument levels and raise vocal tracks at the right times? Can you get the same effect by mixing good levels with the vocals in, and using compression to keep the overall song level when vocals are out?

I'll have to deal with this eventually.

Just to clarify...

The Gain Slider referred to earlier is actually called a Trim Control.

There is also the ability to make manual Clip Gain Envelope adjustments.

These are two separate mixing and editing tools.

As far as mixing the vocalist and Guitarist tracks that contain wildly varying volume levels, a good compressor, properly adjusted during mixdown (not tracking), will be able to raise the quiet sections to a desirable level, while automatically decreasing the level of the loud sections into a more even performance without undesirable artifacts (squashing/pumping/distorting). The compressor will perform much better if you first apply some Clip Gain Envelope corrections to manually adjust the obviously high and unruly transients.

If you're tracking at 24 bit depth, there's no need to track at hot levels or use analog compression that gets printed to disk. Just leave your track faders set to 0.0db and turn down the preamp gain so the peaks don't exceed -6db. We're not tracking to tape here. There's no audible advantage or natural compression from a "saturated" digital track at 24 bit depth like there is when printing to tape. Tracking at 16 bit depth is a different story. Stay as far away from the noise floor as you can without clipping.

I do track vocals and drum overheads with a limiter just for clip protection... but not for compression. When all goes as planned, that limiter is never activated. Then, if a pesky peak sneaks through, the track is not lost due to digital clipping.

I tend to keep mixing levels around -10db to allow room for mastering. I started doing this at the advice of Bob Katz and other mastering engineers, but do it now because I sometimes master my own projects, and I have learned why they ask for the headroom. The keyword here, is want vs. need headroom. Of course they can turn it down for you if they have to. Check it Eric... it's in Bob's book.

If the multi-track mix doesn't sound loud enough to you at -10db, turn up your monitor power amp. It will sound exactly the same to you, and provide plenty of headroom for mastering. The last thing that a mastering engineer wants to do as the FIRST thing he has to do, is decrease the average volume level of your mix. That's an unnecessary DSP operation. The fewer of those the better.

When mastering, the final level is raised to compete with commercial CDs with a limiting threshold of -0.3db. I try hard not to flat top the master, but sometimes the client won't have it any other way.

The loudness war continues.

EDIT: A good range for peaks in a 24 bit depth stereo mix before mastering is -10dBFS to -3dBFS.

EDIT: I corrected the above paragraph on mastering limiting threshold to show a value of -0.3db instead of the -3db value I typed in error. Thanks Eric!

Waltenstein;
When you say gain slider do you mean Trim control? How do you adjust this if you need the mixer slider up to hear the track??

I tend to keep mixing levels around -10db to allow room for mastering. I started doing this at the advice of Bob Katz and other mastering engineers, but do it now because I sometimes master my own projects, and I have learned why they ask for the headroom. The keyword here, is want vs. need headroom. Of course they can turn it down for you if they have to. Check it Eric... it's in Bob's book.

Ah crud.. if Bob say so its probably right. Sorry ohhey.

I'll have to re-read that section, because I'm still not sure what the advantage of losing information prior to mastering is. In the extreme case where the mastering engineer says "i wouldn't change a thing", he in fact will have to add 10db of noise floor to normalize the recording. That seems pretty ugly.

Eric

When mastering, the final level is raised to compete with commercial CDs with a limiting threshold of -3db. I try hard not to flat top the master, but sometimes the client won't have it any other way.

Hey Steve,

Why limit yourself to -3db, rather than -.1 db when mastering? Seems like thats a lot of room you can use to avoid the "flat tops"

Eric

In 24-bit, I track somewhat hot, -3db to -1db peaks. I play/sing a bit before recording to see where the peaks are. ESPECIALLY if I'm going to really chunk a palmed guitar. That's the recording phase.

Then, I pull the volume down so the peaks are around -9 or -10. I do this for each track. This way, the Master should be out of the red when you play the entire song. Now, I mix. Move stuff up, move it back, etc. etc. Once I have everything mixed together how I like it, with fx, eq, compression, etc., and the Master is still under 0db(hopefully) I adjust the volume of each track to bring it up pretty close to 0db, usually moving sliders in .5db increments. So, after I have it mixed, "turned up", and ready for export, I export it to a stereo wav file and open up Soundforge. Throw the wavehammer on it, and now it's CD quality LOUD. The smooth compression setting in WH hasn't disappointed me yet. YMMV.

I'm at work right now, where is that Trim/gain control? I never see it and have to adjust everything with the fader. Is it "hidden" somewhere? THANKS!

ORIGINAL: jazzbass12

Waltenstein;
When you say gain slider do you mean Trim control? How do you adjust this if you need the mixer slider up to hear the track??

Yes, I mean the Trim control. Not sure what you mean with the rest of your question though. What is a "mixer slider"? If you're saying that your tracks are too low of a volume to hear, my suggestion is to record them hotter via manipulation of your recording hardware during tracking. If that still doesn't do it then try turning the volume on your monitors up!

ORIGINAL: Saxon1066

Here's something I always wanted to know: on commercial rock recordings, I've noticed (I think) that when the vocals come in, the levels of all instruments drop back into the background. Ten, during instrumental passages, like between verses, the instruments come back to the forefront.

How is this achieved? Do the mixers use automation (or in the old days, a steady hand) to drop the instrument levels and raise vocal tracks at the right times? Can you get the same effect by mixing good levels with the vocals in, and using compression to keep the overall song level when vocals are out?

I'll have to deal with this eventually.

Usually, they're mixed that way on purpose. indeed, listen close and you'll hear that the mix is always changing, drums come up slightly during flashy fills, the bass drops back when the string swells build up, the guitar riff leads off right out in front, then gets pulled back on the slider and with a hint of reverb as the it repeats through the prechorus, etc. etc.

You can also use a compressor with a sidechain so that compression is triggered on the instruments whenever the vocal comes up. this is called ducking and is used widely in radio voiceovers, paging systems that also play background music, and stuff like that. For an actual song, though, most mixers like the improved control of actually mixing it on the board, just the way they want. Cheers.

and PS- good of you to notice that. there's a lot to be said for asking the right questions.

ORIGINAL: yep

ORIGINAL: Saxon1066

Here's something I always wanted to know: on commercial rock recordings, I've noticed (I think) that when the vocals come in, the levels of all instruments drop back into the background. Ten, during instrumental passages, like between verses, the instruments come back to the forefront.

How is this achieved? Do the mixers use automation (or in the old days, a steady hand) to drop the instrument levels and raise vocal tracks at the right times? Can you get the same effect by mixing good levels with the vocals in, and using compression to keep the overall song level when vocals are out?

Usually, they're mixed that way on purpose. indeed, listen close and you'll hear that the mix is always changing, drums come up slightly during flashy fills, the bass drops back when the string swells build up, the guitar riff leads off right out in front, then gets pulled back on the slider and with a hint of reverb as the it repeats through the prechorus, etc. etc.

You can also use a compressor with a sidechain so that compression is triggered on the instruments whenever the vocal comes up. this is called ducking and is used widely in radio voiceovers, paging systems that also play background music, and stuff like that. For an actual song, though, most mixers like the improved control of actually mixing it on the board, just the way they want. Cheers.

and PS- good of you to notice that. there's a lot to be said for asking the right questions.

This is getting more complicated than a newbie could have imagined. The final mixdown itself, then, is like conducting an orchestra, bringing up the string section, then quieting it down, so the piano can roll over it, then a nod to the percussion to blast in, then everybody down to pianissimo. Thank Sonarus, God of Sound, for automation.

Is there any compressor plugin than does this "ducking"? If you duck it yourself (quack), isn't it painstaking work to get it to around the same total loudness when the vocals are in and out?

ORIGINAL: ebinary
Hey Steve,

Why limit yourself to -3db, rather than -.1 db when mastering? Seems like thats a lot of room you can use to avoid the "flat tops"

Eric

Eric,

You are SO RIGHT!

That should have been -0.3db... not -3db! What a difference a decimal point makes!

Well since I started typing this before I realized why you were questioning it, here's why I limit to -0.3db instead of -0.1db...

From the Waves L2 Ultramaximizer manual:
"For commercial Audio CD, some mastering engineers recommend a ceiling of
–0.3 dB to be safe from distortion due to playback system deficiencies." - Page 18
----
Personal advise to me from Bob Katz on my "24/96 Question" topic on the Mastering WebBoard:
"Using the L2 and the Weiss, I have not gotten bitten with audible problems with the L2 ceiling set to (minus) 0.3. This seems to prevent 99.9% of intersample peaks in typical material from overloading the output of the SRC." - 12/21/03
----
My own personal experience:
I have found that limiting to -0.1db at sample rates greater than 44.1khz many times causes overs in CD Architect PRIOR to sample rate conversion. Yet, if I convert to 44.1khz before loading the file into CD Achitect, there are no overs.

I could understand overs resulting from disturbances or slight changes during sample rate conversion, but -0.1db in Sonar should be -0.1db in any other audio application supporting the same sample rate... right?

Wrong!

I actually contacted Peter Haller at Sony Digital Media and described the problem. He had me send him 2 second clips of each example. Peter had the samples analyzed and there were in fact overs that could be detected on both the 24/96 file as well as the 16/44.1 file. However, in the 16/44.1 file, they were "hidden due to either metering conventions, internal limiting, or requantization".

To correct the problem, the analyst recommended that I "prep the tracks with more headroom (-1.5 dB or so) and add gain in CDA if desired/necessary." - 12/26/03
----

The Moral of This Story:
Not all playback systems and peak meters are created equal.

These are the reasons I set my mastering peak limiter to -0.3db.

ORIGINAL: ebinary
I'll have to re-read that section, because I'm still not sure what the advantage of losing information prior to mastering is. In the extreme case where the mastering engineer says "i wouldn't change a thing", he in fact will have to add 10db of noise floor to normalize the recording. That seems pretty ugly.

Eric

That's actually a good point Eric, so I went back to re-read it myself. Here's the quote:
----
Peak Level Practice for Good 24-bit Recording
"...there is a lot of room at the bottom. and you won't lose any dynamic range if you peak to -3dBFS or even as low as -10dBFS: you'll end up with a cleaner recording. ... A digital mix mix that peaks to -3 dBFS or lower makes it easier to equalize and otherwise process without needing an extra stage of attenuation in the mastering." - Bob Katz, Mastering Audio - the art and the science, Page 65
----

So the acceptable range for peaks in the stereo mix before mastering is -10dBFS to -3dBFS.

ORIGINAL: Saxon1066
This is getting more complicated than a newbie could have imagined. The final mixdown itself, then, is like conducting an orchestra

Exactly! and it's not really THAT laborious, and it's actually pretty fun and extremely rewarding when you nail it. Plenty of demo mixes basically take a "set it and forget it" approach. Nothing wrong with that, but it tends to sound rather dull, and fails to showcase bits that are sometimes the coolest part- i.e. that wicked bass fill is buried in the mix until you turn it up so it pushes everything else out of the way. solution: just turn up the bass on the fill! Or another "static mix" classic: The kick-ass guitar riff that sounds so awesome the first time it kicks in starts to turn pretty mind-numbing by the twentieth time it comes around at the exact same level. So you move it in and out, ever so slightly, and, viola! The listener keeps shifting focus to other instruments, and then gets to hear the guitar riff for the first time all over again, instead of for the tenth.

The thing of it is, when you play live, this stuff is happening anyway. There's seperate sources for every sound, for the guitar, the keys, the hat, the bass, the other guitar, the vocals, the snare, the kick drum, the other other guitar, and all of these things are individual sources coming from human musicians all listening to each other and playing off of each other in dynamic, subtle, exciting ways. But then you go to record it, and you only got two speakers to get all that wild energy to come out of, and everybody's playing one at a time with a click in the headphones, and nervous about their sound and focusing on not making mistakes, and all that stuff, and it ends up a totally different vibe from those beery, sweaty, bright lights, dark corners, dancing people nights in the local club. So the trick is to find creative ways to capture and/or fake that energy and vitality that happens in the room when it's really hot.

Is there any compressor plugin than does this "ducking"? If you duck it yourself (quack), isn't it painstaking work to get it to around the same total loudness when the vocals are in and out?

Plenty of compressors and gates have a side chain. You can run a search on "side chain" or "sidechain" on these forums to find detailed techniques. If you use busses and groups, and stick with a pretty similar set of changes for every time the lead vocals kick in, it's not that tough. And when you think about it, how long does it take to track four or five instruments for a three minute song, presuming you have the whole arrangement, solos, intros, etc. already worked out? Two days? Three? Seven? Isn't it worth it to spend an extra few hours mixing it right? And to tell the truth, if you track four or five songs first and then mix them all after, the first mix usually takes anywhere from a few hours to a day, and then the subsequent ones kind of fall into place in an hour or two apiece. Try it, you'll like it. Cheers.

Thanks for the insights, yep. Actually, sending the rhythm section to a separate bus that drops down a bit during vocals won't be that tough at all (universal bussing is one of the coolest Sonar features). And within that submix, it won't be too hard to push up a cool drum fill for a few seconds.

Waltenstein:
What I ment was the fader control. If I read your post correctly, you had said to adjust you trim (Gain) first then you faders.
So how do you adjust the trim if you have the faders down? Do you adjust all the trims so they so 0?
I might sound ingnorant but, I am just trying to understand the concept.
I have never used the trim controls before.
Thanks

That's actually a good point Eric, so I went back to re-read it myself. Here's the quote:
----
Peak Level Practice for Good 24-bit Recording
"...there is a lot of room at the bottom. and you won't lose any dynamic range if you peak to -3dBFS or even as low as -10dBFS: you'll end up with a cleaner recording. ... A digital mix mix that peaks to -3 dBFS or lower makes it easier to equalize and otherwise process without needing an extra stage of attenuation in the mastering." - Bob Katz, Mastering Audio - the art and the science, Page 65
----

So the acceptable range for peaks in the stereo mix before mastering is -10dBFS to -3dBFS.

SteveD,

Ah, that makes sense in 24-bit land, where you have enough headroom to accomodate a Mardi Gras mask. My original comment was to squeeze right up to -.1db (ok now modified to -.3db), only if you record in 16-bit. Phew!... glad I don't have to question the gospel of the big Bobcat.

And my issue of raising the noise floor does not exist when mastering a 16bit product from a 24 bit source - you'll just be be attenuating less.

Eric

ORIGINAL: jazzbass12

Waltenstein:
What I ment was the fader control. If I read your post correctly, you had said to adjust you trim (Gain) first then you faders.
So how do you adjust the trim if you have the faders down? Do you adjust all the trims so they so 0?
I might sound ingnorant but, I am just trying to understand the concept.
I have never used the trim controls before.
Thanks

I adjust all the trims so they get close to 0db (usually around -2db or so, give or take) before any processing. A simple way to do this is to turn on the metering options "Show Numeric Peak Values" in addition to "Playback Meter Options -> Peak", and "Playback Meter Options -> Pre Fader", then listen to your whole project end to end. On each track there will be a number over to the right of your M-S-R buttons, indicating the peak level you achieved during playback. Now you can set your trim to offset this and give your WAV the desired nominal level. For example, on track 1 your Numeric Peak Value shows "-8.4", so you'd set your Trim control to "+6.4", giving a net of around -2db.

Note that this is just one way to do things. I'm not saying you can't use the trim controls in other ways. I've just found that if all my track bin effects operate on signals that are basically the same level, they will react uniformly to the tweaks that I make when manipulating them. It also makes using Volume controls more uniform across your tracks if you know you're getting around -2db out of every effects bin. You can go straight down the line and set your levels based on the type of track and your desired output level, rather than having to listen again to figure out how loud the thing was recorded.

This actually touches on something I've been meaning to address for a while now - gain structure. I'll try to explain some basics of signal flow and gain structure for you, but this also brings up some questions of my own about how Sonar handles certain things. Just for the record, I'm ignoring the 16/24 debate for now because proper gain structure is relative to a given system's available headroom.

In general, signal flow in Sonar emulates flow through a traditional console. Each channel/bus has an input and and output gain adjustment. On most consoles, the input is the first knob on the channel strip, and is usually labeled "Input Trim". Sonar adopted this convention, so that's the Trim slider. Again, on a traditional console, the output level after passing through the channel strip is set using the fader at the bottom (we're ignoring more complicated mix architecture like matrix mixing, etc.). Same thing in Sonar - the Volume control/fader controls the signal's output level, after channel processing.

The standard convention for gain structure is to use your input to set optimum peak levels, and your output to set mix levels. Translated to Sonar, in practice this means that when recording, set your input levels using your preamps (or whatever you're using to get signal into the CPU) and set your monitoring levels using either the faders in Sonar (if you're using input monitoring) or in your hardware's output section. When mixing, the Trim control and clip gain envelopes adjust the signal level before any other processing. The Volume fader adjusts the signal level after it has been processed (EQ, inserted FX).

Here's where it can get complicated and absolute numbers matter. If you have a broad-spectrum input signal that peaks at -3db, and you use the channel EQ (for example) to boost 500Hz by 6db, you will obviously clip the output. There are ways to compensate for this (turning down the input or cutting frequencies instead of boosting for example), but you can see why it's important to leave headroom on a given track. This is also why mastering engineers like to see peaks between -10db and -6db on raw mixes.

Now, when you're sending a bunch of signals from different outputs (channels) to a single input (a bus), you have to pay attention to gain structure here as well - and each subsequent stage, for that matter. This is called summing. Here is a quote from the Midas XL250 manual (one of the industry-standard big-ass monitor consoles):
"When two coherent test signals are summed together the resultant signal is 6db higher than the inputs...this pattern continues thus: 4 inputs give +12, 8 inputs give +18, 16 inputs give +24.... Fortunately, real audio signals are not at all coherent and are either transient...or partially cancel in the summing process due to phase and frequency differences. Music instrument material typically mixes to give a 6 to 9db increase in level when 48 inputs are summed together."
So with that in mind, you can see why it's important to have headroom in your mix process as well as your individual tracks. It doesn't matter whether you're using 16 or 24 bit: two coherent signals, each peaking at -6db, will sum together at 0db, regardless of bit-rate.

The thing to remember is that clipping in an analog console is tolerable (within reason), but clipping at any stage in a digital system is absolutely hideous and has to be avoided. Keeping the proper gain structure set up all the way through the signal path will ensure you won't have serious problems controlling your final mix levels.

Which brings me to my question: does anybody know the internal structure of the buses in Sonar (bit-rate)? According to traditional analog summing laws, I should be adjusting a given bus's input Trim to compensate for the incoming signals. Some consoles (like the Soundcraft K2 and MH4) reduce the gain of their summing amplifiers by 6db to deal with this, so the "input trim" is under the hood, so-to-speak. Protools handles summing by having a 48-bit bus architecture. So, does Sonar pass signals from the channels to their destination at the project's bit rate or at 32-bit floating point? Or what?

Bill, I hope I've been helpful as far as mix/record level management. If you have any questions, let me know.... Thanks in advance to anyone who knows the deal with the bus architecture.....

Conrad St. Clair
www.kicksville.com

kicksville-

thanks for the lesson! I have learned a lot from your post! even more proof of how excellent this forum can be!

I think he meant -0.3

errrr never mind... it was already addressed... I just didn't read far down enough...

It's done usually via a master limiter on the final output stage. Since the vocals are usually 'out front' in the mix, when the come in, it pushes all other signals back in the mix. It can also be achieved through side-chain limiting/compression.

ORIGINAL: Saxon1066

Here's something I always wanted to know: on commercial rock recordings, I've noticed (I think) that when the vocals come in, the levels of all instruments drop back into the background. Ten, during instrumental passages, like between verses, the instruments come back to the forefront.

How is this achieved? Do the mixers use automation (or in the old days, a steady hand) to drop the instrument levels and raise vocal tracks at the right times? Can you get the same effect by mixing good levels with the vocals in, and using compression to keep the overall song level when vocals are out?

I'll have to deal with this eventually.

Several different philosophies here, then, ba_midi: 1) use automation to drop the accompaniment when vocals come in; 2) use master limiter to push rhythm back from vocals; 3) use side-chain limiting/compressing. I guess I'll try them all. I have used #2 on a Roland VS1880 with fair results. Automation is much easier in Sonar, so I'll try #1 also. Proabably a combination is best.

Kicksville: that was very informative, a good short lesson for newbies. I'm pretty confident I know what to do now. Thanks to all you guys.

ORIGINAL: kicksville
Which brings me to my question: does anybody know the internal structure of the buses in Sonar (bit-rate)? According to traditional analog summing laws, I should be adjusting a given bus's input Trim to compensate for the incoming signals. Some consoles (like the Soundcraft K2 and MH4) reduce the gain of their summing amplifiers by 6db to deal with this, so the "input trim" is under the hood, so-to-speak. Protools handles summing by having a 48-bit bus architecture. So, does Sonar pass signals from the channels to their destination at the project's bit rate or at 32-bit floating point? Or what?

Conrad,

I don't know anything about Sonar's internal bus structure, but I'm curious about your statement regarding the Pro Tools 48-bit bus architecture. How does that alone address summing?

Great post BTW.

I wasn't giving you my philosopy. My philosophy is to get good levels from start to finish, and mix with taste :) (Sometimes easier said than done).

The sound of vocals pushing back the track is common in Radio becuase they use heavy limiting/compression to squeeze signal into the airwaves. However, a good recording shouldn't follow that paradigm or it will just get squashed more.

My 'general' approach is to get decent levels on a per track basis, but also levels which fit the project during recording and mix. What I mean is this:

Let's say you are recording a bunch of tracks - guitars, drums, plugins, whatever. I tend to get levels during the 'record' stage which will already accomodate the final stage (mix). This is not an absolute approach, but a default approach for me. Each track needs its own attention (ie, levels, eq), but overall I feel that the levels set during recording will have an end-result quality. So if you approach everything as "get it as hot as you can", then you can never go back. I don't subscribe to the "get it as hot as you can" paradigm. I think the final sound depends a lot on the input sound. So if something is too hot, it may not be eq'able or compressable to a point where it actually fits in the overal mix of a music project well.

This is HIGHLY subjective stuff - as I'm sure you realize by now. So find what works for you. That's always the best approach.

I do, however, almost always have some sort of limiter or compressor (usually limiter though) on the 'master bus' for a variety of reasons (and speaker protection, grin).

ORIGINAL: Saxon1066

Several different philosophies here, then, ba_midi: 1) use automation to drop the accompaniment when vocals come in; 2) use master limiter to push rhythm back from vocals; 3) use side-chain limiting/compressing. I guess I'll try them all. I have used #2 on a Roland VS1880 with fair results. Automation is much easier in Sonar, so I'll try #1 also. Proabably a combination is best.

Kicksville: that was very informative, a good short lesson for newbies. I'm pretty confident I know what to do now. Thanks to all you guys.

Hi Steve,

This is from Digi's website:

"it's worth noting that Pro Tools uses a 48-bit mixing bus, allowing 128 channels to be summed at maximum gain with no clipping at one extreme, and the ability to pull any or all of the channels' faders to lower than -80 dB and still keep the full 24-bit signal in the mix. "

The whole article is here: http://www.digidesign.com/digizine/archive/digizine_february04/techtalk/

It has some great info on bit rate vs. headroom, and is one of the things that made me start wondering about how Sonar handles its buses. The other was moving back and forth between our Midas and Soundcraft consoles at work - the Midas has no reduction at the summing amps, the Soundcraft has a -6db cut. The Midas rationale is good because you actually know where your levels are at all stages, but you also have to pay more attention to the gain structure downstream from your channels than with the Soundcraft method. Either way works, it's just nice to know which method is being used. Same with Sonar - I won't have clipping if I follow procedure, but I would like to know what is going on under the hood

I hope this answers your question, and thanks....

Conrad
www.kicksville.com

So, does Sonar pass signals from the channels to their destination at the project's bit rate or at 32-bit floating point?

Sonar does all intermediate calculations with 32bit accumulators, so no clipping occurs if you "overdrive"your channels and then reduce the gain prior to the IO device. Well, within 8 additional bits of headroom, that is.

Eric

great - thank you very much. I had a feeling this was the case, but now it's confirmed.....

Conrad
www.kicksville.com

ORIGINAL: jazzbass12

Waltenstein:
What I ment was the fader control. If I read your post correctly, you had said to adjust you trim (Gain) first then you faders.
So how do you adjust the trim if you have the faders down? Do you adjust all the trims so they so 0?
I might sound ingnorant but, I am just trying to understand the concept.
I have never used the trim controls before.
Thanks

Before beginning to mix, play back your tracks with all faders pull completely down. One-by-one push each fader to maximum (+10); adjust that tracks trim control so that the playback meter reads your desired setting (-1); pull down that track's fader and move to the next one. When you've completed every track, you can move all faders up to nominal level to hear all tracks played back with equal volume.

Note the trim setting on each one (they'll differ depending on how they were recorded and the sound of the instrument/voice). Trim adjusment allows you to minimize the fader distance between tracks for mixing...making it easier to control (ride) the mix. Also, at higher fader positions fader movement is more precise than coarser movements in the lower positions, i.e., at the high fader position you can make refined adjustments. Volume balancing is a different process, but good volume balancing begins with gain leveling via trim adjustment.

I just discovered something about my new MOTU 828 MKII... I haven't used it yet, but am getting familiar with the manual. One part says I can toggle between +4 db and -10 db. So In order to leave some headroom in Sonar, I'd assume then I should always record with the setting toggled at -10db? Then I can adjust the gain per track in my MOTU 828, and make sure in Sonar it doesn't exceed -6db's? Does that sound like a plan? :)

Dave

ORIGINAL: billc
Before beginning to mix, play back your tracks with all faders pull completely down. One-by-one push each fader to maximum (+10); adjust that tracks trim control so that the playback meter reads your desired setting (-1); pull down that track's fader and move to the next one. When you've completed every track, you can move all faders up to nominal level to hear all tracks played back with equal volume.

Note the trim setting on each one (they'll differ depending on how they were recorded and the sound of the instrument/voice). Trim adjusment allows you to minimize the fader distance between tracks for mixing...making it easier to control (ride) the mix. Also, at higher fader positions fader movement is more precise than coarser movements in the lower positions, i.e., at the high fader position you can make refined adjustments. Volume balancing is a different process, but good volume balancing begins with gain leveling via trim adjustment.

Hey Bill,

Good tip. Thanks.

Looks like you'll be in my neck of the woods in June.

Maybe I'll stop by the Sands to catch the show and say hey.

ORIGINAL: RRabbi

I just discovered something about my new MOTU 828 MKII... I haven't used it yet, but am getting familiar with the manual. One part says I can toggle between +4 db and -10 db. So In order to leave some headroom in Sonar, I'd assume then I should always record with the setting toggled at -10db?

+4 dB and -10dB are measurements of signal strength that determine whether a device is using balanced or unbalanced connections/cables. Most quality processors will also have a switch that allows the selection of one of the two modes. You should always use balanced connections/settings when possible. If you're connecting with an XLR cable the connection is balanced, so choose +4dB. 1/4 connections can also be balanced, but require TRS (tip-ring-sleeve) plugs on the ends of the cables. Regular 1/4 cables are unbalanced due to having only two wires--balanced cables utilize a third wire that cancels noise, which is the main benefit of using balanced connections.

Hey Bill,

Good tip. Thanks.

Looks like you'll be in my neck of the woods in June.

Maybe I'll stop by the Sands to catch the show and say hey.


Steve, be sure stop by to say hello if you do make the show.

The +4/-10 thing is for connecting your 828 to different kinds of equipment. +4dbu is the industry standard for connecting pro-level gear (usually with balanced lines), and -10dbV is for consumer electronics, turntables, and the like (usually unbalanced). For the record, dbu, dbV and db are all different measurements - here's a good explanation of the differences: http://www.jimprice.com/prosound/db.htm

The idea is to make sure everything in your audio chain is referencing the same absolute levels - two pieces of gear connected at +4dbu will show 0db at the same level. If you're using your stereo to listen to your mixes, your 828's outputs should be set to -10dbV; if you're using pro or semi-pro gear, set your 828 to +4dbu. A good rule of thumb is that if the gear in question uses RCA connectors, it should be set to -10dbV. Gear with XLR or TRS connectors is almost always +4dbu. The +4dbu/-10dbV switch is not a way to address mix/record levels. Keep in mind that dbFS is the scale we usually mean around here when we say "you need to keep your levels at -10db" - dbFS refers to digital 0 (decibels Full-Scale) and not to a measurement of voltage (dbu and dbV)....

Conrad
www.kicksville.com

The +4dbu/-10dbV switch is not a way to address mix/record levels

Everything kicksville says is true, but if your source is at full volume and you are still not getting a strong enough signal from it (meaning you can't hit peaks near -6db and higher in Sonar), you can switch that input to -10. That will effectively boost the analog gain on the input.

Eric

ORIGINAL: kicksville
Keep in mind that dbFS is the scale we usually mean around here when we say "you need to keep your levels at -10db" - dbFS refers to digital 0 (decibels Full-Scale) and not to a measurement of voltage (dbu and dbV)....

Conrad
www.kicksville.com

Last time I checked... A picture was still worth a thousand words...

Waltenstein

Waltenstein
Thanks for the info. very informative. Here is what I tried, does this sound correct?:
First I set all my faders to 0. I play the song and monitor my levels- Then I adjust all my trims to a peak of 0.
I figure you need to calibrate the faders first that way you are setting the trims from a flat reference. is this the same theory as "Playback Meter Options -> Pre Fader", ?

sweet....

Conrad
www.kicksville.com

Not necesssarily.

The gain has to do with the type of inputs you'll be using. I use -10 since I'm using line inputs to the sound card and line outs.

And while I do tend to keep 'individual' tracks around -6, that's not written in stone. I really depends on the signal - and the sound you're looking for.

I think you just have to experiment with what works best for your environment - and of course for the type of music you're doing at the time.

ORIGINAL: RRabbi

I just discovered something about my new MOTU 828 MKII... I haven't used it yet, but am getting familiar with the manual. One part says I can toggle between +4 db and -10 db. So In order to leave some headroom in Sonar, I'd assume then I should always record with the setting toggled at -10db? Then I can adjust the gain per track in my MOTU 828, and make sure in Sonar it doesn't exceed -6db's? Does that sound like a plan? :)

Dave

ORIGINAL: kicksville

This actually touches on something I've been meaning to address for a while now - gain structure. I'll try to explain some basics of signal flow and gain structure for you, but this also brings up some questions of my own about how Sonar handles certain things. Just for the record, I'm ignoring the 16/24 debate for now because proper gain structure is relative to a given system's available headroom.

In general, signal flow in Sonar emulates flow through a traditional console. Each channel/bus has an input and and output gain adjustment. On most consoles, the input is the first knob on the channel strip, and is usually labeled "Input Trim". Sonar adopted this convention, so that's the Trim slider. Again, on a traditional console, the output level after passing through the channel strip is set using the fader at the bottom (we're ignoring more complicated mix architecture like matrix mixing, etc.). Same thing in Sonar - the Volume control/fader controls the signal's output level, after channel processing.

The standard convention for gain structure is to use your input to set optimum peak levels, and your output to set mix levels. Translated to Sonar, in practice this means that when recording, set your input levels using your preamps (or whatever you're using to get signal into the CPU) and set your monitoring levels using either the faders in Sonar (if you're using input monitoring) or in your hardware's output section. When mixing, the Trim control and clip gain envelopes adjust the signal level before any other processing. The Volume fader adjusts the signal level after it has been processed (EQ, inserted FX).

Here's where it can get complicated and absolute numbers matter. If you have a broad-spectrum input signal that peaks at -3db, and you use the channel EQ (for example) to boost 500Hz by 6db, you will obviously clip the output. There are ways to compensate for this (turning down the input or cutting frequencies instead of boosting for example), but you can see why it's important to leave headroom on a given track. This is also why mastering engineers like to see peaks between -10db and -6db on raw mixes.

Now, when you're sending a bunch of signals from different outputs (channels) to a single input (a bus), you have to pay attention to gain structure here as well - and each subsequent stage, for that matter. This is called summing. Here is a quote from the Midas XL250 manual (one of the industry-standard big-ass monitor consoles):
"When two coherent test signals are summed together the resultant signal is 6db higher than the inputs...this pattern continues thus: 4 inputs give +12, 8 inputs give +18, 16 inputs give +24.... Fortunately, real audio signals are not at all coherent and are either transient...or partially cancel in the summing process due to phase and frequency differences. Music instrument material typically mixes to give a 6 to 9db increase in level when 48 inputs are summed together."
So with that in mind, you can see why it's important to have headroom in your mix process as well as your individual tracks. It doesn't matter whether you're using 16 or 24 bit: two coherent signals, each peaking at -6db, will sum together at 0db, regardless of bit-rate.

The thing to remember is that clipping in an analog console is tolerable (within reason), but clipping at any stage in a digital system is absolutely hideous and has to be avoided. Keeping the proper gain structure set up all the way through the signal path will ensure you won't have serious problems controlling your final mix levels.

Which brings me to my question: does anybody know the internal structure of the buses in Sonar (bit-rate)? According to traditional analog summing laws, I should be adjusting a given bus's input Trim to compensate for the incoming signals. Some consoles (like the Soundcraft K2 and MH4) reduce the gain of their summing amplifiers by 6db to deal with this, so the "input trim" is under the hood, so-to-speak. Protools handles summing by having a 48-bit bus architecture. So, does Sonar pass signals from the channels to their destination at the project's bit rate or at 32-bit floating point? Or what?

Bill, I hope I've been helpful as far as mix/record level management. If you have any questions, let me know.... Thanks in advance to anyone who knows the deal with the bus architecture.....

Conrad St. Clair
www.kicksville.com

What a great thread...it really should be a sticky!

This is what the forums be all about.

I have always assumed that PA9's and Sonar's internal bit-rate was 32, and by reckoning that when he meters displayed 0db it was really -3db (got this value by comparing the values between the meters in Sonar to the ones in the Delta control panel). Following these guidlines I do not see clipping in the wave display. I avoid compression in the recordings (except in the case of our kick drum, where we have no choice due to the extreme dynamic differences in our drummer's playing - hey man we love ya'). When we run 16 bit our SNR comes out to a reliable 80 db, and going 24 bit I get about 98 db. Only process in the final mix, and I do the mastering myself (tightwad on a shoestring budget). My normalization level is set to 98.6 db. I may be totally wrong but I'm a dancin' fool.....

I am consistently amazed by three things on these forums:

1. The obsessive, almost panicky compulsion of so many people who clearly are very new to all this to make sure that they know the "correct" technical minutae.

2. the sheer amount of gross misinformation and/or speculative or philisophical opinion that is presented as fact.

2. The patience and generosity of the people who are in the know in their willingness to answer these questions and actually clear up the urgent questions like whether your masters should be at -.03 dB or -.01 on the meters.

If I were going to try and post something constructive to this thread, I would probably say something like (this is me, personally):

a. Sonar has plenty of internal headroom. If you get a good mix and it clips on the main outs occaisionally, just turn down the mains bus in sonar and you should be fine.
b. For best results, get your mixes to sound good and then check the main outs meter and adjust the master fader as necessary to "print" them with peaks somewhere close to, but below, 0dB on the meter.
c. If you are worried about maxing out Sonar's internal headroom because you routinely run all of your tracks deep in the red, then stop doing that.
d. If you really want your CD to match the sound and level of "professional" commercial CDs, send it to a good mastering house once you've got the best mix you can get on your own.

But instead, I think I'll sit this one out.

cheers.

I smell something burning, and I don't thiink it's my CDR drive....

Ok, if it's not to late here, can someone check me through this once. (I fear i might just be not connecting a dot somewhere..

billc
before beginning to mix, play back your tracks with all faders pull completely down. One-by-one push each fader to maximum (+10); adjust that tracks trim control so that the playback meter reads your desired setting (-1); pull down that track's fader and move to the next one. When you've completed every track, you can move all faders up to nominal level to hear all tracks played back with equal volume.

Waltenstein Studios
I adjust all the trims so they get close to 0db (usually around -2db or so, give or take) before any processing. A simple way to do this is to turn on the metering options "Show Numeric Peak Values" in addition to "Playback Meter Options -> Peak", and "Playback Meter Options -> Pre Fader", then listen to your whole project end to end. On each track there will be a number over to the right of your M-S-R buttons, indicating the peak level you achieved during playback. Now you can set your trim to offset this and give your WAV the desired nominal level. For example, on track 1 your Numeric Peak Value shows "-8.4", so you'd set your Trim control to "+6.4", giving a net of around -2db.

This set-up step results in your raw mix-tracks all aligned to around -1 or two peak. (?)
If the 'nominal' target on the tracks faders is zero', and depending on the number of tracks (buts lets assume we have a dozen or more), it would seem that you have added an extra step that would cause these to combine at the master in excess of zero.
(As opposed to coming from my admittedly seat of the pants approach and experience..).. if I record most of my tracks at generally conservative levels (let's say loose combination from -18 to -3) and leave the trim at zero, (and again, depending on how many tracks) I'm still apt to have to pull many of them back at the fader for the mix to come in or around the zero range.

Isn't that one of the points of recording at modest A/D levels (building toward some natural system headroom)?
Don't we want to avoid any gain point being outside of zero' that we can? (Just like why we wouldn't arbitrarily normalize' our tracks?)
Wayne

SteveD,

Good point!!

So can you shed some light on the difefrence of dbRMS versus dbFS.

It can be confusing for a newbie to mixing if you will.

I hear folks say get the recording to about -0.3 db

Then I hear get it to around -3.0

Is this at the mastering stage or what?

Right now most of my recording prior to mastering read from -12 to -6, when I view them in Adobe Aditions 1.5. From here I find myself adjusting so that it ends up as close to -.01 as possible, lately if I get it close -3.0 I stop.

Which is correct?

What am I looking at in Adobe Audition 1.5 rms or fs? I see the number along the side that seem to represent db. But while playing back there is a meter at the bottom that seems to show the recording staying just below -3.0 yet it still sounds full!

My recordings don't sound bad at all but then again I'm listening on a pair of Mackie HR24's. Occaisionally I go back and forth between the headphones and a small pair of multi-media speakers to test how it will sound on various systems.

This same exact conversation is going on in another thread titled Final Mix RMS levels. Is this not the same issue? Help me out with this guys!

Per Track Pre-Mix

1. -27dB Starting Volume -Phase Correct (that means check your MS and correct for problems.

2. Bandwidth not to exceed 20hz-20khz, trim everything under or over that before you start hearing really hideous side effects from DAC correction & bandwidth accumulation.

3. Bit Depth level set to target medium and conformed (usually 16 bit depth for CD otherwise 24 for DVD-A, though only a few dozen houses are set up for it in the states currently)