Garritan Instruments and Mixing Levels

[RichardK]

I'm interested in people's thoughts about optimum mixing levels with 100% Garritan instrument pieces. I have all the libraries now and they differ markedly in the default volume settings when loading into the Aria player (though the Aria fader may be set the same for each). Some have to be adjusted -12dB or more on the Aria player to balance with others. Then if Aria instances are loaded as VST plugins there is the decision of mixing level adjustment in the DAW vs. Aria.

I'm wondering if there are tips or benefits to adjusting levels in Aria vs. the DAW vs. both and is there an output level in the final mix that tends to work best with Garritan instruments (of course depending in the type of music this might vary but in general are their targets that one tries to meet with Garritan instruments to get the best sound?).

Thanks
Rich Kram

[rbowser]

Hello, Richard - It's a really good post, full of good questions. I'll insert my feedback:

I'm interested in people's thoughts about optimum mixing levels with 100% Garritan instrument pieces.

As you say at the end of your post, levels in a mix is partly determined by the style of music. But I'll make it even more general - The "optimum mixing levels" is going to change with every project. It isn't really possible to say "Here are the levels you should use." I'll make it easier though - You just need to start with every instrument playing at a good healthy level, at about the same level as everything else. Then you start pulling things down as you go. Some level changes needed will be easy to determine - For instance there's no way a single clarinet's level could possibly be the same as a bass trombone. So, even though you've started with those two instruments at the same level, you know right away that the two have to be put in a balance that sounds more realistic.

I have all the libraries now and they differ markedly in the default volume settings when loading into the Aria player (though the Aria fader may be set the same for each). Some have to be adjusted -12dB or more on the Aria player to balance with others.

Exactly right. The volume differences are huge between the Libraries. They were developed at different times, sometimes by different teams of people, and each Library is something of an island unto itself, with its volumes not being done in the way of the next Library at all. For instance all of the COMB instruments are much softer than GPO. When I mix those two Libraries, I always have to push the COMB instruments up, and pull GPO's down. Sometimes there are discrepancies inside one Library - like the Viola in GPO jumps out at you, it's so much louder than the other strings.

So that gets back to what I said in the first reply. Load those instruments from different Libraries, and the first task is to even them all out. The row of volume sliders end up looking like Godzilla teeth, zigzagging up and down in order to just get the instruments all at the same volume before you start making the adjustments you need.

Then if Aria instances are loaded as VST plugins there is the decision of mixing level adjustment in the DAW vs. Aria.

I'm wondering if there are tips or benefits to adjusting levels in Aria vs. the DAW vs. both and is there an output level in the final mix that tends to work best with Garritan instruments...

I use Sonar, and like other DAW programs, the MIDI faders on its tracks take control of ARIA's. If you've had the situation where you set levels in ARIA, start your project, and those faders all jump to a different position, that's a case of your program taking control. My advice is once you're in a DAW program, you should stretch out, give yourself room to work, and use all the controls the program has - that as compared to working in the smaller and more limited ARIA mixer. I set my ARIA levels through Sonar's sliders, and also the pans.

It sounds like you're wanting to keep your projects in the MIDI realm, getting levels and all that the way you want, and then exporting the final product. I think most people work that way nowadays. But I use Sonar even more thoroughly - I bounce all my MIDI tracks to audio before I start mixing. I do everything I can with MIDI, then bounce - then mute, archive, and hide the MIDI tracks since I don't need them anymore. I turn off the synths. Now I'm working with just audio, as if I've recorded tracks of live musicians, and I'm mixing in the time honored audio engineer way. Even though I can do a lot with my MIDI tracks, I found a long time ago that I can do So much more working with the solid audio files - I can polish up the sound of a piece much more than if I stay with just MIDI. There's a much wider dynamic range of volume available, for instance. Maybe I went as loud as I could on some tracks - but in audio there's still headroom of quite a few DBs. Those forte passages can get Much more forte.

And so forth. It takes more time, some people find it frustrating - but I just have to work with audio.

So for me there's no question that the actual detailed mixing, with all those automation envelopes on my audio tracks, is done in the DAW. The sound coming out of ARIA is just the beginning.

If you don't bounce to audio, there's still a lot you can do outside of ARIA, there in your recording program.

No matter what kind of piece of music it is, or how the mix is done, when you're looking at your mixer and working on that final mix - you just want to keep the sum total of all the sound, which is going to your Master - keep it so it's staying below 0 DB, certainly not into the red. It may be a wee bit weaker than you want - but that's when you go to:

Your audio editing programming. You really need one. Audacity is fine, it's free, and that's why it's a standard for a lot of people. All audio editors do pretty much the same thing. That's where you import your 2-track master that you mixed in your DAW program. That's where you can look at the waveform of the entire thing, and bring it up to the peak level it should be - just a few hairs under 0 DB. Do that with a compressor limiter on gentle settings. Without going wild with those plugins, you won't hear damage to your music, but you'll maximize its impact. Normalize up to around 98% - that's the way to go up to a high ceiling without going over into the red.

If you don't want to use an audio editor, then it's a bit harder - You have to get the final level you want there in your DAW program. And that can be done - You can use a compressor there too- you can ride the gain etc. It's just that it's best to simply mix, keep the volume healthy but not to maximum- and then later focus on the big picture of what that mix is sounding like.

AAAAANd so forth! lol. - There's some things your post made me think of, some of which I'm sure went into areas you weren't actually asking about. But I've tried to cover both the more basic way of working as well as a bit more advanced one.

To recap, looking at your original questions again - It's not as if working with Garritan Instruments gives you any particularly unique mixing problem to solve. It's just like working with any sound, either from samplers, synths, or physical instruments. Get your volumes up to a healthy level - Start with everything at about the same level, adjust as needed - then do your fine tune mixing - AND that mixing really Has to involve volume automation on your tracks, not just MIDI volume data.

Enough for now!

Randy

[RichardK]

Randy,
Thank you so much for this lucid response! I do in fact use Wavelab after the fact, but I don't bounce tracks to audio - I need to start thinking about that and try your other suggestions as well.

Rich

[rbowser]

Randy,
Thank you so much for this lucid response! I do in fact use Wavelab after the fact, but I don't bounce tracks to audio - I need to start thinking about that and try your other suggestions as well.

Rich

I let my fingers fly on the keyboard last night when I wrote my reply, Richard - Could have used some editing! But I'm glad you found some useful things in what I wrote, thanks for letting me know.

Good deal that you use Wavelab. As I said in my novel earlier, it's much better to not force your mix into being The Master straight from your recording program.

Important caveat about bouncing to audio: It's something I used to recommend often - But last year there were a couple of people at the old Garritan Forum who got intrigued by what I was saying, tried it out, and got very frustrated because it seemed like too much more to learn. Their experiences made me back off from recommending it so often. On the other hand, at least one person tried it out, was able to get good results, and now he doesn't want to go back to just MIDI because he discovered what I was saying, that more control over the end result is potentially available that way.

I'm just saying - Some people seem to "get it" more quickly than others, when they try out mixing with audio. For some, it's just a headache. So - just wanted you to know, it's a process that doesn't seem suited for everyone.

Randy

[RichardK]

Randy,
My goal is to create the best music I can. The more I learn the better off I will be. Actually I used to bounce to audio now and then when I had a slower computer that could not keep up with the samples and plugins - and at some point anyone creating huge scores like I sometimes do will have this issue even with the fastest of computers. It also might let me better utilize some plugins that take more processing power than to try and do them all in line like I am now. It's actually a great suggestion in my mind. Frankly I don't do it out of laziness - which is a bad mindset.

Thanks Again,
Rich

[rbowser]

Randy,
My goal is to create the best music I can. The more I learn the better off I will be. Actually I used to bounce to audio now and then when I had a slower computer that could not keep up with the samples and plugins - and at some point anyone creating huge scores like I sometimes do will have this issue even with the fastest of computers. It also might let me better utilize some plugins that take more processing power than to try and do them all in line like I am now. It's actually a great suggestion in my mind. Frankly I don't do it out of laziness - which is a bad mindset.

Thanks Again,
Rich

That's an excellent goal, Richard. Some people, mostly notation users, are strictly focused on composing, and don't care to spend time also on producing the best possible renderings of their works. There's nothing "wrong" with their approach, and it's easy to sympathize with their complaint that they just don't have time or patience to spend yet more time at their computers trying to be sound engineers.

Then, other people enjoy not only composing, but also taking more full advantage of the sound production facilities available on our modern computers. You seem to be in that category.

I also used to bounce tracks on an older computer because I couldn't play all the tracks I needed in real time. I would work up the arrangement for a group of maybe 6 instruments, bounce them to audio, archive the MIDI tracks, and then go on with my arrangement, developing the rest of the arrangement while playing back the audio and creating more MIDI tracks. Then, as you're saying, with our newer, more powerful computers, bouncing isn't as much of a necessity. Within reason, we can play back full orchestra scores which are MIDI only.

But I want to add another caveat to what I was saying about all this yesterday: There are many proficient professionals who don't bounce their tracks. They do their arrangements, keep their projects in that very flexible MIDI realm - flexible since changes can easily be made up until the last minute - and they produce great recordings.

So, it's not as if rendering MIDi to audio is a necessity. I've continued bouncing because even though I do everything I can with my MIDI tracks, I'm not as satisfied with the renderings straight from MIDI as I am with the ones I do from audio tracks. That's basically what I said yesterday.

Here's a screenshot that helps illustrate what I'm talking about. It's of a mono audio track with a volume envelope that was recorded in real time and then adjusted by moving the nodes more precisely where I wanted them. With the instant visual feedback of seeing exactly how the sound is fluctuating in that segment, I was able to improve the dynamics already in the track done with MIDI work. You can see at the beginning that I wanted to boost that first section of notes, then I wanted to pull back the next one. Throughout the rest, there was room for improvement, as there always is. With that kind of detail on every track, it's important to understand that the automation is mostly done in real time - It would be very laborious to do all that envelope work by hand, but of course there's always some tweaking to be done by hand.



Something else about mixing a project like this is that it gives me assurance that the project can be archived in a safe, solid way. I have a folder with the project file which of course contains the MIDI data, and the audio sub-folder of all the individual tracks. That project can be re-visited, re-mixed, without relying only on MIDI and a particular line up of soft synths which may have changed, may no longer function on an updated computer, may have been discontinued - etc. I like knowing I have a solid archive, much like studios having their old multi-track masters to dust off when they want to do an improved re-mix of an old song.

And so forth.

Richard, I'm pulling out part of that quote from you again, because you bring up something else I want to comment on:

...It also might let me better utilize some plugins that take more processing power than to try and do them all in line like I am now...

I think you're talking about printing an effect on an audio track so you don't need to run the plugin in real time. You're of course right that it's a way to conserve CPU power - but I highly recommend you don't do that. That's destructive editing, resulting in wet tracks that can't be changed without the hassle of deleting them, turning on the MIDI track and synth again, and rendering a new copy. You really want to keep all your project audio tracks dry, and run their FX plugins in real time - Much more flexible, and it's the only way I ever work.

I don't generally use a lot of plugins, and so that makes it more possible to work as I do. On some projects, I will still use just one reverb plugin, strapped to a bus, and then the instrument tracks have their Send knobs set to how much reverb I want - the basic, classic way of using reverb. Sometimes I'll use multiple reverb buses, but rarely more than maybe 4. EQ is already built in to each Sonar track, so while I'll do some EQ work on tracks, that only takes a tiny amount of power. Otherwise, that's all I use on a lot of things. Once in awhile there will be a delay plugin on one track, or a chorus plugin - something of that sort, but only rarely.

If you're using a lot of plugins, then, yes, they could add up to being quite a drain. I would encourage you to continue running them in real time as much as you can though.

Have a good day, and I hope you have time for working on some music!

Randy

[RichardK]

Randy. Thanks Again.
If you don't mind, one more question. Lets say I have a final mix out of Cubase that is -7dB when I bring it into Waves. I bring this into Waves and scale it to 0dB in most cases? I've found that for some things -3, -2 or so sounds better to my ears. I normally scale to a level instead of just raising everything to a level to maintain more dynamic headroom. Is this more or less what you are talking about?
Thanks
Rich

Thanks
Rich

[rbowser]

Randy. Thanks Again.
If you don't mind, one more question. Lets say I have a final mix out of Cubase that is -7dB when I bring it into Waves. I bring this into Waves and scale it to 0dB in most cases? I've found that for some things -3, -2 or so sounds better to my ears. I normally scale to a level instead of just raising everything to a level to maintain more dynamic headroom. Is this more or less what you are talking about?
Thanks
Rich

Yes, it's only a generalized thing to say you should aim for just a fraction under 0 DBs when doing your mastering. Pieces that are mostly of gentle, quiet nature wouldn't sound right mastered at the usual -0.3 DBs, but not really because any damage is done to the sound, but because in proportion to Other recordings, it will seem too loud.

The thing about volume levels is that are always relative to Something. The overall volume of a recording doesn't exist by itself, on its own, but in comparison to:

-- Whatever the listener heard before it.

-- Whatever level the listener usually uses when playing music.

Like a lot of people, I keep my playback system generally at one volume. I know when I play professional recordings, it's going to be at the level I want. When I listen to home studio recordings, it's always a crap shoot, with a lot of recordings being way too soft. Finale renderings are often extremely low in volume when users don't use a sound editor to bring the volume up to what it should be. So I have to turn the volume up to hear a lot of that. But we want to aim for a volume where a lot of volume knob twiddling isn't called for.

Ever look at the waveform of an entire album? It's an interesting thing to see. The loud pieces will all be up there close to 0, the soft pieces will be down lower, like the -3 DB you mentioned. But initially, those individual recordings for the album were all mastered at optimum full volume for the best sound quality. So those soft pieces have loud masters, but then their volume is mastered for the album to be much lower so the contrast between pieces will be appropriate. And of course there was no reduction in sound quality because the volume was lowered, not brought up. See?

So what's safe to do is to master your softer pieces at a lower volume so they'll play back at an appropriate volume for people who optimized their playback systems.

Remember that when you make MP3 copies, that process raises the volume a bit, so you want to anticipate that, and that's why you never go for absolute 0 DB. If you make your MP3, check it, and see that it peaks out, one thing you can do is use your audio editor's pencil tool to re-draw those red peaks down to where they don't get flat tops - and you can do that without doing damage. Use a "detect peaks" kind of plugin to see wherever you've accidentally gone over the top.

And, one more note, of course pieces with really large dynamic ranges, like most classical music -those will have long sections which are way low, with only the forte passages reaching up there to 0. A big dynamic range is theoretically nice, but works better in a live performance than in a recording. You don't want people having to adjust their volume knobs while a piece is in progress - If it gets too soft, and they're having to turn it up, then you've kept the dynamic range too large in the recording.

Randy

[RichardK]

Randy,
Wow. Thanks for taking the time to explain all this. It both confirms that some of what I am doing is valid and also gives me a lot of good advice to make things better!

For example, you get some very clean and bright sounding IO examples mixed very well (my first attempt was muddy and levels were wrong). I think a lot of what you are saying here is at the heart of that. I'm heading back to the studio and going to apply some of this.

Thanks Again!
Rich

[Credo]

Don't forget that the Aria Player effects (Convolution and Ambiance) can have a pretty big impact on the perceived levels of the patches. I.E. If a channel is using such an effect and you've turned up the send for the instrument in the Aria Player then it's going to add the processed version of that sound in addition to the dry mix. It also seems to me that the processed send is going to output over audio streams 1/2...even if you've assigned the dry version of that voice to some other outputs.

In my experience, with the Ultimate Collection (as of Sept. 2014) the various collections of samples/presets are very consistent and quite well calibrated when you play them dry (no post processing effects).

When I go into 'mix-down' mode I'm pretty quick to use the Multi-Output version of Aria Player, disable all the post processing effects in Aria Player itself, make sure the effect send pots are all turned to zero, start with a dry mix, and then use the various VST pulgins and Aux Busses in my DAW to go after the type of mix, blend, positioning, and ambiance I'm seeking. With the multi-output version of Aria, you can isolate each of the 16 channels to different audio streams and hand the mixing and effect processing over to the DAW.

Don't get me wrong, the post processing built right into Aria Player is quite good and highly capable! I just find it easier to work my various VST instruments through the DAW's mixer and effects consoles than I do going into each instance of Aria (and other VST instruments) to tweak out a mix. In using the power of my DAW, I can put mixing tasks in the same place...and easily assign MIDI remote controllers to every little aspect of post signal processing...and even automate the most minute of post-processing details on the fly.

When I do use Aria's Convolution and Ambiance settings in the Multi-Output version...I tend to reserve outputs 1/2 strictly for hosting the effects, and I don't assign any actual channel instruments to 1/2. In short...I treat channels 1/2 like an independent FX bus.