Evening Track Volume Levels on a CD

[efiebke]

Hello -

For the first time, ever, I'm attempting to write a "Suite" which will be my 2008 Holiday CD for friends and family. I have plans for four orchestral compositions. Two are written and sequenced, so far. (I should have started writing this project much earlier! YIKES!)

Anyway. . . After finishing the second orchestral piece (of the, hopefully, four pieces), I decided to burn a CD of the two finished compositions. I was surprised to find that the overall level of the second piece was more than a bit softer in volume than the first. I am using the EXACT same orchestral instruments for the entire project. During the sequencing process, I use the same volume and panning settings for each of the instruments. I am also using the same reverb plug-ins and settings for the entire project as well. So, with this sequencing strategy in place, I am left wondering WHY there exists differing volume levels between piece #1 and piece #2. The loud sections of piece #2 is softer in volume (with the similar instrumentation) than the loud sections of piece #1.

This is the first time that I am creating such a project. Usually when I create these Holiday CD gifts, there is a mixture of jazz and orchestral styles of music. This would, understandably, present quite a challange in creating a final CD mix. However, I started the 2008 Holiday CD project with hopes that the mix-down process would be much easier because of the same instrumentation and plug-in settings. I guess it ain't easier as I thought. (I am no expert in sound engineering and mixing down, by the way.)

So. . . here are the big questions. For those of you who compose lots and lots of orchestral compositions using orchestral libraries like our beloved GPO, what is YOUR working strategy in sequencing to ensure that one piece is basically the same level in volume as the next? How do you keep the "soft section" of your Piece #1 similar in volume to the "soft section" of Piece #2? Do you find yourself with a similar issue when using the SAME instrumentation from one piece to the next? If so, what do you do to keep all of the same orchestrations in proper level with each other?

Thank you, and happy composing, folks! :)

Ted

[rbowser-]

Hiya, Ted - I'm glad I have at least a few minutes to give you some ideas on your question. I'm not sure if I've ever had the opportunity to pass on some hopefully helpful tips to you before.

In brief - Orchestral projects so easily become fairly complex to work on, simply because of the number of instruments involved. As home studio musicians, we're not only dealing with the performance of each instrument, but also with all of the technical considerations of trying to get recordings of those tracks sounding as good as we know how.

So much for stating the obvious - In-between the lines there is the implication that since these projects are complex, any number of unknown issues can come up which yield unpredictable results, such as the ones you mention here.

Why is one recording unexpectedly louder than the other which was put together the same way? Hard to tell. There has to be something different that you're not noticing, in a mixer's slider perhaps. Maybe you recorded some automation without being aware of it - that's happened to me.

But here's really the crux of what I wanted to say to you in reply - I feel it's not practical to rely completely on sequencing/notating to yield the aural results we want. The first step is the sequencing, but it's really only the beginning. That's working in fairly broad strokes. The largest portion of time on a project is in the editing of the sequence, and then in the audio editing of the recordings made from the MIDI tracks.

I think you're probably already doing a lot of "tweaking" in the MIDI realm. But, for instance, you're limiting the potential dynamic range of a recording if you rely totally on MIDI volume control. On a routine basis I will have worked hard to get great dynamic contrasts via MIDI, but invariably once I'm working with the audio files, I'm punching up the dynamics even more, automating the audio faders to swoop up on the highs, and swoop down even farther on the lows - etc. I will have done the best I can with balancing the various instruments, but also discover while mixing the audio, that I can do a Lot better in fine-tuning that relative volume balance.

So much can be done to polish up raw MIDI tracks through audio editing - That's why DAWs are so great to work with - there are basically two platforms to work in, first the MIDI, then the audio, complete with virtual mixers, audio plugins et al.

It's in working on a project once its in the audio realm where I also even out the relative volume levels between pieces. After doing everything I can with Sonar's mixer and other audio tools, I take the resulting 2-track mix down and bring it into Sound Forge, a program where final mastering can be done. There in Forge, using its audio editing tools, I make sure that the loudest sections of a piece are reaching right up to "0" DB. With two tracks open for comparison, it's easy to even the files out so they both have the exact same decibel level at their peaks - You can see that visually as well as hear it.

When dealing with a larger number of tracks, it's easy to string them together, compare their wave profiles - In some cases, a mellower piece will need its peak value to be lower, so that there's a variety in the flow of the collection, but not grossly so.

But you're wanting these two pieces you mentioned to have the same dynamic spread, so that one piece's highs are at the same volume as another's---What I've described is the only way to do it. Relying solely on MIDI control won't yield you the amount of accurate control you need for what you're asking.

--SO - If you don't have a dedicated mastering program like Forge, you can do much the same thing still inside a DAW like Sonar. In both cases, it's a matter of matching peak volumes on a 2-track master - not when a piece is still unmixed.

-------Hoping this more helpful than confusing--gotta go for now, best of luck, Ted!

Randy B.

[GrahamKeitch]

Hello Ted,

I can sympathise perhaps more than I can help - but I'm very familiar with the problem. It's one that a lot of us face when putting a multi-movement piece together.

As Randy has mentioned, there are many sliders that need checking for consistency - depending on your workflow. I work straight out of Finale (no DAW) but I can still get caught out if base velocity, human playback, studio view mixer vol and the expression libraries (and probably a few other things) are out of whack. So, I either work with one thundering great project that contains several movements or, I save the first project as a template, delete all the notes and call it movement two - or whatever. That way, it re-uses all the same settings.

There is also the opportunity to normalise volume when you burn the CD - but I never do that because to me, that's interfering with the performance. But in the past I have had occasion to change the overall volume of a track at burn time to bring it to the level I intended. I went through a phase of taking key velocity right down on the strings to get a smoother, warmer sound and then eveything else had to come down to the same level - and I'd end up with an over quiet track.

Good luck,

Graham

[rbowser-]

Hello again, Ted -

I hope the responses you've had here are helpful.

I want to add to something Graham said in response:

"...There is also the opportunity to normalise volume when you burn the CD - but I never do that because to me, that's interfering with the performance. But in the past I have had occasion to change the overall volume of a track at burn time to bring it to the level I intended..."

Normalization is a standard mastering process which is a bit misunderstood. Begging Graham's pardon, I have to say that it doesn't interfere with a performance, as he suggested. All it does is bring the entire audio file up to a level where its peaks are at zero DB. That's a very desirable thing - all dynamics are left intact and exactly as before, but now the listener doesn't have to turn up his volume knob just to make your track in equal volume to what he heard before.

The last step in mastering a track in fact is normalization. Before that, one may have used a compressor to make the extremes in volume a bit less than they were originally. I need to compress a file often, at least to some degree, when I've gotten carried away getting these dramatic contrasts in volumes.

One of the great benefits of compressing and normalizing is that they makee one's tracks more in the mode of what a listener is accustomed to hearing. Everything we hear in film and television, for instance, has been compressed, usually quite heavily. We're used to hearing soft passages still played at a volume where we can easily hear it. The dynamics still exist, just on a smaller, more subtle scale than if audio files hadn't been compressed at all.

Everything you've ever heard from me on The Forum has been mastered in Sound Forge, and that process has included both soft compression and normalization.

Graham's second sentence, about adjusting the volume of a track in order to bring it up to the level he intended - even if it's done manually, that has the same effect as automatically done compression and/or normalization. The goal in all cases is to bring the peaks of a track right on up to zero DB. If you leave a track lower than zero (prime volume)--it has to always be for a very good reason, like wanting it to contrast dramatically with the track heard before or after it.

Randy B.

[efiebke]

Hello Randy and Graham -

First, I want to thank you for taking the time to respond. Your thoughtful and thorough input is sincerely appreciated.

I am, of course, sequencing this project. I'm just confortable with this method of working with computer-based instrumentation. I use as many of the CC's needed to help create a "realistic" sound for each instrument or group of instruments. So, Randy, you are so correct that one must be mindful of all of the individual settings for each instrument. Although the main volume (CC 7) and panning (CC 10) are the same for each instrument for this particular project, some of the other CC's may not be the same. For instance, the "extressive" controller to control the volume of the "Strad" might very well be different between the two projects. This is also true for the modulation wheel (CC 1) which controls real-time volume for many of the other GPO/JABB instruments. Especially for the CC 1, I attempt to keep the quiet, loud and very loud at fairly consistant levels. But even this may vary a bit from project to project (not to mention section to section) which may cause the overall volume differences between the two compositions. :eek:

Randy. You're also so correct with regards to the editing process. I can spend HOURS trying to mix with instrumentation so that it sounds at least adequate. I actually don't mind doing all of this work. I must confess, though, that I spent less time on this newer composition. I can see where I just might need to edit some of the velocities values for the orchestral percussion. This does take time. UGH!!

With regards to normalization. Although I used to do this all of the time, I usually don't do this as much anymore, to be honest. I tend to mix the piece so that the loudest parts are around -0.3 db or so. This leaves me room for future mixing if needed. I read this particular strategy once on some music-based bulletin board somewhere on the internet. But I guess one can say that my "normalization" is around -0.3 db. LOL! ~|

Oops. . . got to get going!! I'm come back to this! Got to get going, though! . . .

Ted

[keithjfuller]

I think whats happening is that you are normalizing all your tracks. What then happens is this, you had a piece that was quiet throughout so it normalized loud. Then you had a piece that was very dynamic (some soft parts and then loud, and maybe only one loud part), so after normalizing that piece it came out really quiet. Make sense?

[GrahamKeitch]

Hello again, Ted -

I hope the responses you've had here are helpful.

I want to add to something Graham said in response:

"...There is also the opportunity to normalise volume when you burn the CD - but I never do that because to me, that's interfering with the performance. But in the past I have had occasion to change the overall volume of a track at burn time to bring it to the level I intended..."

Normalization is a standard mastering process which is a bit misunderstood. Begging Graham's pardon, I have to say that it doesn't interfere with a performance, as he suggested. All it does is bring the entire audio file up to a level where its peaks are at zero DB. That's a very desirable thing - all dynamics are left intact and exactly as before, but now the listener doesn't have to turn up his volume knob just to make your track in equal volume to what he heard before.

The last step in mastering a track in fact is normalization. Before that, one may have used a compressor to make the extremes in volume a bit less than they were originally. I need to compress a file often, at least to some degree, when I've gotten carried away getting these dramatic contrasts in volumes.

One of the great benefits of compressing and normalizing is that they makee one's tracks more in the mode of what a listener is accustomed to hearing. Everything we hear in film and television, for instance, has been compressed, usually quite heavily. We're used to hearing soft passages still played at a volume where we can easily hear it. The dynamics still exist, just on a smaller, more subtle scale than if audio files hadn't been compressed at all.

Everything you've ever heard from me on The Forum has been mastered in Sound Forge, and that process has included both soft compression and normalization.

Graham's second sentence, about adjusting the volume of a track in order to bring it up to the level he intended - even if it's done manually, that has the same effect as automatically done compression and/or normalization. The goal in all cases is to bring the peaks of a track right on up to zero DB. If you leave a track lower than zero (prime volume)--it has to always be for a very good reason, like wanting it to contrast dramatically with the track heard before or after it.

Randy B.

Hi Randy, thanks for the clarification.

The point about peaking at zero DB is also a useful point. I'm going to recheck some of my recordings because I've never been too sure about the optimum levels.

Regards, Graham

[rbowser-]

HI y'all - I like the -0.3 DB approach, Ted. On the occassions when I use either a normalizer or "volume maximizer," I actually have those set at
-0.3. It was an over-simplification to just say Zero DB earlier. Long ago I saw the same advice that you referred to, and it's good advice. It avoids having those momentary in-the-red moments which can slip by if you literally bring a track up to zero.

Interesting though, brief in-the-reds aren't always heard. I've seen advice from people who believe that Louder Is Better (all modern pop producers) - who say they don't worry about those brief moments. They just avoid the definitely ugly digital over-load sound we all have heard and hate. Those are more sustained periods of being in the red.

Sounds to me like you're doing great, Ted. I really feel the evening out that you want is best controlled in the audio world, especially in a dedicated mastering program where you get a nice big, detailed view of the wave forms and at a glance you can see what's happening with volume fluctuations - and I'm talking about 2-track mix-downs.

And so on.

Stay busy!
Randy

[Hannes_F]

Ted,

this is exactly the difference between mixing and mastering.

Mixing is more or less the art to balance the lines within one track. You do your thing with cc's, bounce it or not, apply more or less reverb, etc. until the track itself sounds fine.

Mastering is the art to balance different tracks for a CD, film or whatever. Now the track stands in a context to other tracks and the whole CD in a context to other CDs. And you don't want it to stick out in regards to other tracks to either side.

For mastering orchestral tracks you check the track-to-track-consistency of three aspects:

- stereo image
- overall EQ
- volume

Therefore your tools are a stereo imager, the most colour-less parametric EQ you have (preferably a minimum phase EQ) and the gain.

And preferably good listening conditions like a monitor system in a treated mixing room. In case you don't have this try to listen to your music under different conditions and don't skip the car since it may be better than thought. Check the bass with a headphone now and then.

Then, if adjusting the gain is not enough you can go further with a limiter or, depending on the music, with a mastering compressor.

BTW professional mastering should be about 80 bucks per track, I have never used this as a service so far but am sure it is worth its money. It begins with the room which is very neutral and expensive (if you go to a real professional mastering studio, that is).

[Hannes_F]

I should add that these are typical levels of music production that I use:

- Recording peaks at - 18 dB each track (in case of 24 bit recording)
- The mix peaks at - 6 dB
- The mastered track will peak at - 0.3 dB for loud tracks

If you are working with the notation output of GPO then you should check whether your notation host works in 16 or 24 bit. 16 bit tracks should peak at say - 6 to - 12 dB to begin with. Overall it is better to keep a project in 24 bits because the reverb tails are much better.

Two very important things:

- If you bounce something with reverb, bounce it in realtime. The difference to a fast bounce is very high (at least for convolution reverb).

- The very last step for going from 24 bit to 16 bit for the CD should include a dithering process.

Cheers
-H

[Reegs]

Hi Ted,

Perhaps the level change you heard might be due to the density of the pieces? More instruments playing will produce a louder volume.

If I remember correctly, you're on a mac. Some options for 2-track mastering are Adobe Audition (which does more than you really need), and Bias Peak. Peak LE is the little brother of Peak Pro and has all the functions you'll need to master a CD. Its drawbacks are that it only loads 1 VST plugin at a time and that all of the options in Peak Pro are listed in Peak LE but greyed out. Those are in Version5 anyway--6 might have changed things.

Also I believe built into Logic is a Match EQ plugin--very useful in the mastering stage for getting consistent sound.

If your track levels are high enough throughout, you likely won't hear it appreciably, but you should use the highest quality dither you have. My favorites are MBIT (in Izotope Ozone) and Pow-r 3. Dan will say that you don't need to dither at all.

I only differ in Hannes' advice in that I like to peak at -12dB when recording at 24-bit. :)

As people have alluded, normalizing to precisely 0.0dB is a bad idea. Never normalize to 0.0dB. Always leave a little headroom for the transients in between samples and for possible further processing. Also remember that volume in dB's is a logarithmic scale.

Reegs

[efiebke]

Hello - Thank you, again, for your time in responding to my question. :)

Reegs -


Perhaps the level change you heard might be due to the density of the pieces? More instruments playing will produce a louder volume.


I think that this might be a significant part of my "problem". I'm relatively new to the world of orchestration. Although I am using the same total instrumentation for this particular musical project, the second piece has different combinations of instruments from the first one. Definately, the density of sound from the instrumental combinations are different between the two pieces, even at the "louder" sections.

Regarding normalization

I used to use the normalize plug-ins all the time until I read somewhere that it might not be good to do so. At present, I do not use any normalization plug-ins for the more orchestral compositions. Instead, I try to mix the individual and grouped instruments so that the loudest portions of the compositions are no louder than -0.3db. I do this either by bringing the phaders down (individual or grouped) or tweeking the CC #1 values (or which ever controller "makes" the instrument louder or softer). In all honestly, I haven't used a normalization plug-in in years. I also tend not to use any compression plug-ins. Actually, I avoid using the compression plug-ins for any of the orchestral pieces, again relying on tweeking the CC values or the mixer-phaders.


Hannes_F

I appreciate your input on the difference between mixing and mastering. Your suggestions on mastering orchestral pieces is very helpful! I own Peak Pro with all of its bells and whistles (sp?). I know that its power is in mastering. Unfortunately, I haven't a clue on how to master the tracks together. Your posts here have begun to answer a lot of mastering-based questions for me. THANK YOU! :)

Again, EVERYONE, thank you for taking the time to address my concerns. Your thoughtful responses are most appreciated! :)

Happy composing (and sequencing, and mixing and mastering) folks! :D

Ted

[P.T.]

Matching levels of the various sections and pieces should be left for the end.
Just work on getting a good mix for each piece and then set the relative levels of each section and piece so that they sound right.

You can also use volume automation to control the volume for desired effect.

Once the relative levels are right then you can normalize to around -0.2 or -0.3 and then burn the CD.

[buckshead]

Make a copy of the first sequence, then add the second sequence to the end with a suitable gap = one sequence containing two tracks add sugar and stir slowly over a low heat. Check the overall mix and save to CD. It cant be different its now all one track.