Mastering Orchestral Tracks

[0x000000FF]

Hi, everyone. So I'll start by saying that I do know about the pinned production thread and I've taken a look at it more than once. I'm hoping for some specific advice, I guess.

I've been doing some reading and experimentation, and I just have no clue how to approach mastering and production. I find that I really just don't know where to start, even after I've identified that a track sounds okay in one environment but terrible in another.

Here are two examples. They're both raw, and I'm sure even the initial mixing could use some serious work too. The first is soft and the second is loud, but I find that both are muddy. Where would you start with these? What obvious problems should I attack first? Can anyone offer any advice in the context of these examples? I find that both can sound pretty horrible on the right (wrong) speakers. I've been trying a lot of orchestral tracks, and the muddy results are typically the same.

I've tried basic equalization and applying filters to instruments (in hopes of making things sound less muddy), but the results are usually a very "airy" mix and I always seem to end up with some over-represented frequencies that kill the ear (I suppose that's just trial and error). I'm guessing I should take more time to learn about parametric EQ to help remedy that.

Just a note, but I've actually reduced the headroom on those tracks since I knew I was not going to process them further, so I bumped up the gain a bit. Without any additional boost or compression, I aim for between -4 to -2 peak dB.

Also (and I hope I didn't miss this from the pinned thread), I don't own decent speakers/monitors. I was thinking of picking up some good headphones. Any recommendations?

Thanks a lot! I appreciate the help.

[Rozovian]

If you want something in between your mud and your airy ones, just ease up on the filter strengths a bit. Actually, if you're not talking about eq, you might be going about it the wrong way, as both LP and HP filters cut out everything beyond the threshold (gradually depending on the filter strength).

Most ppl will say you shouldn't EQ an orchestra, but unless you're working with the real thing you'll probably have to anyway, in order to get your fake instruments to sound better together. Different make means different sound, and while real orchestra instruments are built to work well together, not every soundfont out there will sound right with all the others, same with VST and everything else. So unless you mean EQ when you say filter, change tool and use eq instead. Just be subtle.

Also, carve in the reverbs. Not all instruments need all their reverb, some of them might do well with just half their mid-range reverb, while others won't sound right with too crisp highs in the reverb. A little EQ on the right track's reverb might go a long way to clean things up.

Orchestral isn't my forte, so take this all with that disclaimer in mind.

[Moseph]

What orchestra samples are you using? Is this a combination of free Soundfonts? A comprehensive orchestral library? I ask, because related to Rozovian's discussion of EQ, usually things from a comprehensive library (or a decent comprehensive library, at least) don't need a lot of EQ to fit together well, whereas if you're combining multiple sources, you may need heavier EQ.

The problems I hear between the two examples are actually pretty different. In the Cave Story one, I think it largely has to do with an unrealistic and dead-sounding acoustic space, which is mostly a reverb issue. I'd recommend a convolution reverb unit if you aren't using one. It makes a huge difference for orchestra music. Reverberate is an extremely good cheap one, or try Reverberate LE or SIR as free solutions. Also, there's not a lot of performance subtlety here, which is exacerbating the unrealism of the acoustic space. All of the notes sound like they're played at the same velocity all the time, which makes the music sound extremely static. What's available to you in your samples' velocity layers -- if the samples have velocity layers at all -- may restrain what you can do about this, but you should at least try to get some differentiation of level going on so you can shape your patterns and phrases into something that sounds like a real performance instead of a machine.

The Chrono Trigger example sounds very thin to me, not at all like the other example. This is likely an EQ issue. You're missing a lot of low end -- it sounds like maybe everything below 500 Hz has been cut too much. Generally, I shoot for a feeling of warm fullness in the low end, which is admittedly a vague way to describe the goal, but the low end end needs to feel like it's present to you in some way, like you can reach out and touch it, if that makes any sense, and that's definitely not happening here. The Cave Story example comes closer to it.

EDIT: Also, yes, you need some decent headphones or speakers. It's possible that the mud you're hearing is partially caused by what you're listening on. I own some old desktop speakers that make well-mixed things sound muddy and boomy. I know I've mixed something right when I've gotten it to sound awful on those speakers . The Chrono Trigger example on my mixing headphones (AKG K702) is way more tinny than muddy, whereas it actually sounds pretty good on the terrible desktop speakers. There have been several threads about buying headphones in the past, so you should be able to do a forum search to find info on what to buy. If you're looking to spend under $100, the AKG K240 seems to have a lot of fans.

[0x000000FF]

Thanks a lot for the great insight! It's really helpful.

Since it seems relevant, I'll mention that I'm (mostly) using EWQLSO, but I've avoided actually customizing the instruments using the VSTi itself and have made most adjustments in the tracker (Renoise). I'm willing to bet that's a problem!

So unless you mean EQ when you say filter, change tool and use eq instead. Just be subtle.

When I said filter I really meant filter (like LP and HP, not EQ). I read some general advise elsewhere on the forums that filters tend to be underused while EQ tends to be overused, so I've tried to make more use of filters than EQs. The filters are pretty light (the cutoffs are quite low for HP and quite high for LP). I have noticed that simply throwing an EQ on master that cuts back some of the mid frequencies seems to sound a bit better.

In the Cave Story one, I think it largely has to do with an unrealistic and dead-sounding acoustic space, which is mostly a reverb issue. I'd recommend a convolution reverb unit if you aren't using one.

I actually don't even know what convolution reverb is! I'll look into the units you recommended. Would I want to dial back the reverb applied by EWQLSO in favor of the convolution reverb?

...you should at least try to get some differentiation of level going on so you can shape your patterns and phrases into something that sounds like a real performance instead of a machine.

Yes, with a few exceptions, I don't change up the velocities much at all. Do you have any guidance on how to approach this? I tried to add a bit of variation to the dynamics at one point, but I wasn't pleased with the results. Maybe I should just pay more attention to this when I'm listening to orchestral music to get the idea.

The Chrono Trigger example sounds very thin to me, not at all like the other example. This is likely an EQ issue. You're missing a lot of low end -- it sounds like maybe everything below 500 Hz has been cut too much.

It looks like I mistakenly cut back on the low frequencies. I pull them back just a bit with an EQ, but I also cut them off with filters on the double basses. Yikes. I'll see if I can warm them up a bit.

I'm curious how people approach applying filters and EQs to tracks. I find it difficult to tell if I've improved or worsened the sound when listening to a single track/instrument/section and adjusting the DSPs. Maybe this is something that will just take time to develop? I'm sure having decent speakers would also make this easier.

This could just be my awful speakers, but I also find that it's terribly easy to exaggerate high frequencies, and cymbals and high frequency percussion sometimes sounds overblown (clipped and compressed) even though I haven't compressed those particular tracks. What kind of EQs or filters should I typically apply to percussion?

If you're looking to spend under $100, the AKG K240 seems to have a lot of fans.

Yep, I'd like to avoid spending too much. These look like a great choice.

Thanks again for the help!

[Nabeel Ansari]

To me it sounds like you're adding filters and EQ to enhance the sound of EVERYTHING, which is not really what you're supposed to do.

You use EQ to cut back frequencies when a specific range is too loud or conflicting with another instrument. For instance,

synth A has mainly mid high frequencies while a synth pad B has lots of mid lows AND mid highs.

If you played them together they would conflict because they're both fighting for the same range. If you used an EQ or filter to cut the mid highs (but not all the way, you still want some room to breathe) on synth pad B, you give synth A more room to cut through the mix.

That's the basic theory behind EQ for mixing. For mastering, if the mix generally is biased towards mid to highs, you can boost the bass, within reason so it still sounds balanced.

Yes I used synthesizers in my example, but this generally goes for anything especially if you're doing orchestral with instruments from different sources. Like they said above, stuff from the same library go fine together, but stuff from different sources don't.

[SonicThHedgog]

Thanks a lot for the great insight! It's really helpful.

Since it seems relevant, I'll mention that I'm (mostly) using EWQLSO, but I've avoided actually customizing the instruments using the VSTi itself and have made most adjustments in the tracker (Renoise). I'm willing to bet that's a problem!

Thanks again for the help!

Convolution reverb is a simulation of reverb, so if you have any kinda vst that can load impulses and you have impulse's givem a spin.

I use VoxengoBoogex to load impulse, and its also as guitar cap but can work dry as well! I would not use the keFIR vst on verb impulse's because its VERY high on cpu(like 2cAther verb , man i like 2c but too much cpu.... ). but both Vst Iisted are free.

Here is what I use http://www.memi.de/echochamber/responses/index.html

[Nabeel Ansari]

I use VoxengoBoogex to load impulse, and its also as guitar cap

What the fruck is a guitar cap?

[SonicThHedgog]

What the fruck is a guitar cap?

Try google sometime

[Nabeel Ansari]

Try google sometime

That's a rock and roll cap.

[SonicThHedgog]

That's a rock and roll cap.

http://imagethumbnails.milo.com/009/676/499/290/9676700_11513499_290.jpg

[AngelCityOutlaw]

My favorite thing about EQ, is watching all the pretty lights of the spectral analysis dance.

[martin.demsky]

one great composer told me this and really it helped me to get more natural strings:

Make sure you use LASS supplied IRs with a good convolution reverb like Wizoo for the Early Reflections. This will place the sound in the correct position. For Late Reflections / Tail use an algorithmic reverb, like Lexicon or Bricasti M7. Another VERY VERY important thing is to make sure you have a subharmonic processor/synthesizer like the DBX 120A. Depending on the pace of your track/mix you might need to use in the end of your signal a very subtle transient designer to make the mix cleaner for other sounds.

Otherwise i always analyse in SoundForge soundtracks i like, and i try to get that sort of sound.

[dannthr]

Since it seems relevant, I'll mention that I'm (mostly) using EWQLSO, but I've avoided actually customizing the instruments using the VSTi itself and have made most adjustments in the tracker (Renoise). I'm willing to bet that's a problem!

I'm not sure if you're at the level where you can start customizing this sort of stuff. Using a tracker may end up being an issue for you, I'm not sure how MIDI versatile Renoise is.

I actually don't even know what convolution reverb is! I'll look into the units you recommended. Would I want to dial back the reverb applied by EWQLSO in favor of the convolution reverb?

Again, EWQLSO has a lot of HALL noise built into the samples, I'm not convinced that this is going to bring more life to your tracks--rather I contend that there's a lot of life already in the samples themselves, and when you get far enough along, you'll realize that this can actually be annoying. What I will say is this: EWQLSO is designed to sound mixed OUT OF THE BOX. It's pre-panned, pre-processed, and generally speaking it can sound quite good with only two skills applied librally: Good writing and good midi execution.

Yes, with a few exceptions, I don't change up the velocities much at all. Do you have any guidance on how to approach this? I tried to add a bit of variation to the dynamics at one point, but I wasn't pleased with the results. Maybe I should just pay more attention to this when I'm listening to orchestral music to get the idea.

Well, not only is it a problem that you don't change up velocities, but that you don't employ EWQLSO's Dynamic CrossFade instruments (that's the dxf or xfd instruments) which allow you to breathe expression and life into your samples to create virtual performances.

I would also say that as it is, your writing could use some help. You're not writing for orchestra, you're writing an electronica track with orchestra samples. This is a problem.

Not sure about your mix? Not sure about your instrumentation? Not sure about your writing?

It's time to engage your analytical listening skills. You should check out the GREAT selection of youtube videos posted by the Berliner Philharmoniker (The Berlin Philharmonic) which has an incredibly detailed sound due to their fairly advanced mic arrangement and some great live mixing--here's their youtube channel: http://www.youtube.com/user/BerlinPhil

Your goal should be to mimic the sound of real live instruments as closely as possible using every MIDI tool at your disposal, and if it's EWQLSO, then your tools are going to be MIDI Velocity, MIDI Pitch wheel, MIDI CC1, MIDI CC11, MIDI CC7 mostly. If you don't know what a MIDI CC is, that's a problem--educate yourself in your DAW.

Here's a good sounding mix: http://www.youtube.com/user/BerlinPhil#p/u/8/Yj-4ew4p8pE

As far as writing goes, that will be a life-long journey if you're up to it.

Here's a good piece to listen/analyze throughout that journey: http://www.youtube.com/watch?v=fZtsUzpRtm4

Good luck and if you want to chat more about this, you can PM me your AIM.

Cheers,

- Dan

[0x000000FF]

Thanks for the insight!

Using a tracker may end up being an issue for you, I'm not sure how MIDI versatile Renoise is.

Do you mean it will be an issue because trackers are generally not suited for this (orchestral music)?

Again, EWQLSO has a lot of HALL noise built into the samples, I'm not convinced that this is going to bring more life to your tracks--rather I contend that there's a lot of life already in the samples themselves ... EWQLSO is designed to sound mixed OUT OF THE BOX.

Yes, I know there's already a lot going on with the EWQLSO samples. When I first tried EWQLSO, I immediately noticed the reverb and panning, for example. I tried playing with different reverb DSPs, and I agree that it's probably best to just work with what's already baked into the samples.

It's pre-panned, pre-processed, and generally speaking it can sound quite good with only two skills applied librally: Good writing and good midi execution.

...

Well, not only is it a problem that you don't change up velocities, but that you don't employ EWQLSO's Dynamic CrossFade instruments (that's the dxf or xfd instruments) which allow you to breathe expression and life into your samples to create virtual performances.

My MIDI controller broke down a while back and I entered notes for these tracks using mostly a computer keyboard and a KORG nanoKEY controller with velocity sensitivity disabled. I plan to go back and re-enter/perform with modulation/pitch wheels and a better Control Change scheme. As such, I've used no dynamic articulations (yet). Other than some velocity changes, the MIDI execution here is utterly flat.

I would also say that as it is, your writing could use some help. You're not writing for orchestra, you're writing an electronica track with orchestra samples. This is a problem.

Can you point out some examples? I feel like this is more true of the Cave Story track than of the Chrono Trigger track, but I'm curious what you think about this. I'd like to know more about how to avoid this problem.

You should check out the GREAT selection of youtube videos posted by the Berliner Philharmoniker (The Berlin Philharmonic) which has an incredibly detailed sound due to their fairly advanced mic arrangement and some great live mixing...

Thanks for the link. I'll check it out.

As far as writing goes, that will be a life-long journey if you're up to it.

That's part of why I'm here. Thanks again for the help! I really appreciate it.