Jump to content

Are the rules for EQing synths different?


144dB

Recommended Posts

  • Members

Hey all,

 

I've read plenty of recording/mixing books, and a very common idea is that if you have the right mic, right mic placement, and the right player/instrument, EQ is largely unnecessary.

 

I'm not sure that applies to synths, or compositions that are synth-heavy. 95% of what I do is keyboard compositions (consisting of ROMplers, digital synths, and soft synths), and given the synthetic nature of the sounds, I find that I have to eliminate more resonances than I would with "real" instruments. I guess I subscribe to the Bruce Swedien adage that says, "I don't care if I have to turn the knob around backwards. If it sounds good, it is good."

 

If something is bothering me with a sound, I always try to address it at the source. I typically drop one of the stereo synth channels and just use a mono track, and sometimes I'll add a short delay to the other side of the stereo spectrum to regain some width. For the sound itself, maybe there is an oscillator in the sound that is too loud relative to the others. Maybe the filter or envelopes need some tweaking. Maybe there is some detuning or chorus buried in the sound that is causing some problems. But once I fix those items, I find that I still have to EQ a lot to get sounds to lay together in a composition. And these EQ tweaks are almost always cuts. I almost never boost a frequency, absent the occasional high shelf to give a part some air or clarity.

 

I have attached some screenshots below from Fabfiler. On the EP track (Roland XV-3080), there were a lot of resonances I didn't like, I just found the problematic frequencies and dipped them out with a high-Q notch.

 

Given your own experience, am I on the right path, or do you find that you don't need much EQ with synths?

 

Thanks in advance.

7842d37ed6fa0172f748d852e5804b83.JPG.c153eef3e2a56eef4ab9bb7eebae839e.JPG

6a37fc34c18c47ad964229d13771e03b.JPG.3f33f7569aa9263d2f4e62cc0e64e674.JPG

7570ce88fc1eb5e376042a4b0fe26e73.JPG.79acf9955d087e1d36f8f80b5ec2f3d6.JPG

Link to comment
Share on other sites

  • CMS Author

What rules? And what synths? You're on the right path, though. Synthesizers tend to extend over a pretty broad range of the audio spectrum, so if they're not the solo instrument, there are a lot of things that they can get in the way of. It looks, from what you're doing with your equalizers, is reducing their energy to make room for something else. And if there's a "rule" about equalizing in a musical context, that's it.

 

The trick, if you haven't discovered it, is turn off the synth and listen for what you can hear better now. If it's something important, find a frequency range that you can cut in the synth that will let what you're missing (or didn't know you were missing) through. And if you're piling synth voices playing the same part on top of one another, unless you've designed the patches yourself with this in mind, there's bound to be some frequencies in common that will build up unnaturally, and that you'll want to tame.

 

But, yeah, you can look at the construction of the patch as your "equalizer." If you're going to be playing a pad under a vocal, then arrange your waveforms and filters so that they aren't covering up the articulation and character of the vocal.

Link to comment
Share on other sites

  • Members

Thanks guys...

 

Today was not a great day in the studio... My new cans are great (KRK KNS 8400s), but they are highlighting just how bad some sounds really are. This is good, because I couldn't hear some of that detail before, but it's also a stark reminder of how much further I have to go, and it can be overwhelming.

 

In the quest for pro quality, mastered sound, I've been working on a very simple EP track. It's just a Roland XV-3080 patch with a bit of UAD reverb and PSP delay. I'm struggling like hell to get it up to standard listening level (which I define as roughly -15dBFS on an RMS basis), without it sounding boomy and resonant. I've tweaked the patch and dipped-out some of problematic frequencies, but when you cut that much out of the sound, it really sucks the gain out of it. Before too long, you have to boost it to high heaven with a limiter, and it raises the noise floor to an audible level.

 

Maybe I need to work on the patch some more. The way I see it, if I can't get a single patch to sound fantastic, it's going to be even more difficult to get a full synth mix to sound fantastic. This is proving to be a challenge...

 

 

Link to comment
Share on other sites

  • Members

I should add that it's not the limiter ruining the sound... It's not even engaging, it's just raising the gain (peaks are still well below 0 dBFS).

 

It's just that the sound becomes really unpleasant at professional gain levels (-15 dBFS to -12 dBFS RMS).

Link to comment
Share on other sites

  • Members

Working with romplers and/or VST sampled instruments, the thing is to just find a really good patch in the first place. I've gone through the drill of struggling with some patch or other, only to have someone come along and say, "geez, that sampled sackbut of yours is really noisy and full of artifacts - here, try this patch I've got" and 90% of my problems go away by just switching to a better sample set. Sure, I might still need to put a lo-pass filter on it or dip out the lo-mids or sparkle up the highs, but there's no upside trying to fix a lousy sample set with drastic EQ or other processing. Just shouldn't have to do that these days.

 

Sampled instruments have gradually gotten better and better over time - stuff that used to sound great some 10-15 years ago...usually there's something newer that blows the older stuff away, and costs a lot less, too. The people creating these things have gotten a lot better at making patches that drop into mixes easier.

 

That XV-3080 of yours is getting pretty long in the tooth - there are a zillion newer VSTs out there that cover all the old rompler ground- check out the electric pianos offered by Waves really cheap.

 

best o luck -

 

nat

Link to comment
Share on other sites

  • Members
Also might check the fx on the 3080. IIRC, those older romplers had a tendency to drench stuff in effects for a "wow" response from browsing gear fanatics at GC, etc. Maybe try just muting or changing the dry/wet mix, see what you get....

 

Also a lot of the presets are designed to sound good by themselves on the floor at a Guitar Center, not necessarily in context with other tracks. I think you'll probably find yourself cutting a lot more than boosting.

 

Link to comment
Share on other sites

  • Members
Also might check the fx on the 3080. IIRC, those older romplers had a tendency to drench stuff in effects for a "wow" response from browsing gear fanatics at GC, etc. Maybe try just muting or changing the dry/wet mix, see what you get....

 

nat

 

 

Yep - I turned off all onboard effects before I started. Unless the effect is subtle or absolutely integral to a sound, I never use the onboard effects.

 

Link to comment
Share on other sites

  • Members
Working with romplers and/or VST sampled instruments, the thing is to just find a really good patch in the first place.

 

Hi Nat. That's a very big part of the challenge. I have quite a few instruments to pick from (both software and hardware), but nothing I have comes close to the character of this sound. Believe me, I've tried. Most of the EPs I have are the typical "dirty" or slightly overdriven Wurlis or Rhodes, and this sound is a blend between a Rhodes and a synth (almost like an MKS-80, an MKS-20, and a real Rhodes all layered together).

 

On the EP front, I have my MP11, my PC361, HALion 5.x, FM8, and my XV-3080. I also have the Korg Legacy Collection, but the M1 and Wavestation are really played-out in this category. As much as I love my PC361, the EPs are really weak. My MP11 has an awesome tremolo'd Rhodes, but it's not the right sound for this simple piece.

 

I'll keep trying. Giving up isn't an option.

Link to comment
Share on other sites

  • Members
The people creating these things have gotten a lot better at making patches that drop into mixes easier.

 

 

I might disagree with you on that one... One reason I like the older instruments is because they tend to sit in a mix well, particularly a busy arrangement. So many of the Motifs, Kronoses, Fantoms, etc. have such big sounds, that they don't lend themselves to being used in an arrangement. This is along the line of what Craig said. The patches are designed to wow on the shop floor, but they are way too big and fat to fit into a mix. Some of those big sounds are indeed fun to play, but when it comes to finishing projects, they never get air time. Some of those old ROMplers with their 12bit and 16bit samples really help glue a mix together. But on the flipside, some of them also have the problems that I'm struggling with right now.

 

That's one thing I really love about Kurzweil. In addition to their ultra-deep programming and customization opportunities, their sounds tend to be really useful. For as little ROM as they use, their orchestral sounds are still phenomenal. Their KB3 organs and VA1 digital oscillators are great too. The pianos and EPs are getting long in the tooth, but they still make a great, great overall instrument. I hope they do a "PC4" or "K3000" in the not too distant future. That PC361 I have is deep.

 

 

Link to comment
Share on other sites

  • Members

Hey guys,

 

I posted some files so you can hear what I'm hearing.

 

Here is the original XV-3080 preset, which has the "Hexa-Chorus" effect engaged. There is no processing on this, other than about 4dB of limiting to bring it up to standard listening levels.

 

https://clyp.it/11lky0nu

 

Here is the same clip, with Hexa Chorus removed, some onboard EQ applied, and some heavy dipping with Fabfilter in Cubase. It was also sent through a Limiter, but with about 10dB of boost to bring it up standard listening levels. The limiter never actually clamped down on the sound, it was just used to raise the gain.

 

https://clyp.it/i4eollqk

 

Finally, here is the EQ'd clip with some PSP42 delay added and a UAD reverb (I don't recall if I used the EMT140 or the Lexicon 224). If you listen close, you can actually hear distortion, which is odd given that the waveform is nowhere near a clipping limit.

 

https://clyp.it/atnxtqzc

 

Some may listen to the first clip and say, "Hmm, it sounds fine". My ears are really sensitive to 250-400 Hz, and it sounds really boxy and unpleasant to me (along with hearing some resonances at higher frequencies).

 

For the second and third EQd versions, it's really odd to me that there is distortion of the sound. The waveforms are nowhere near clipping, and the only conclusion I can draw is that there is too much energy at a mid or low frequency, which is causing certain speakers or headphones to break up. I can hear it clearly, but I'm not sure what's causing it.

 

Any criticism or insight is welcome. I'm still learning, and I'm sure I can learn something from this. Thanks again.

 

 

Link to comment
Share on other sites

  • Members

Well, the basic patch certainly sounds like a classic Roland big,big,big Rhodes with a lot of frequency content low to high and their signature chorus effect. Yes, I totally agree that in almost any mix except the most sparse, that patch would need some thinning out, no question. It doesn't sound boxy to me so much as just uber-lush and fat. As mentioned, really impressive on the showroom floor, but too tubby to just drop into a mix unprocessed.

 

The third clip is noisy even before the playing begins. The stereo effect is mostly gone in versions two and three. If you are running a stereo sound through a mono effect, that might be causing some issues.

 

Please forgive if you already know this stuff, but each of your effects has some sort of input and output level control. Or at least an output control. So including the original patch, there could be as many as ten level knobs at work from the beginning of the signal path to the final output. If you've dinked around with a VAST instrument, you probably already know about chasing down distortion in a complex signal path. Gotta look at each input and output stage methodically to find the culprit.

 

Since the original patch sounds to me the way Roland intended it, my best guess is that the limiter, doing that big boost, is where the distortion begins. In your DAW can you just normalize instead of using the limiter and/or other effects to boost the gain to the listening level you want? That will bring up your original undistorted sound to the volume level you need and your plugins and/or hardware processors won't have to be pushed so hard.

 

It's the old rule - fix things as early in the chain as possible, give all the processing gizmos as light a duty as possible.

 

Unless you can edit your 3080 patch right in the Roland unit to tamp down those lo-mids, you're going to be stuck with EQing out the excess phat. But try to get rid of the necessity of using a limiter or other plugin to do so much boosting - they aren't really made for that sort of duty by and large.

 

best o' luck

 

nat

 

 

 

 

 

 

 

Link to comment
Share on other sites

  • Members

Hi Nat,

 

First off, thanks for listening to the clips and providing advice. I'm learning some new things here, and it's tremendously helpful.

 

When I was using my UAD Precision Limiter to boost the level, I wasn't aware that I was putting a strain on it. Since it wasn't actually limiting or compressing the gain in any way, I thought it was just acting like an adjustable gain boost. In other words, if I used the limiter or if I did a one-time gain uplift in Wavelab, I thought the results would be identical. It never occurred to me that they wouldn't be, given that the plugin wasn't actually compressing the gain.

 

Tonight I took the patch and swapped out the onboard Hexa Chorus for an onboard stereo EQ. I then made some pretty drastic cuts to the sound before it ever hit my converters. To give you an idea of the magnitude, the original patch could easily digitally clip my inputs. After the cuts, the XV could be full bore, and it would only register about -20 to -18 dBFS on Cubase's meters. Again, these were some pretty drastic cuts with a wide 0.5 Q value.

 

I then did some more cuts with Fabfilter to remove a few resonant peaks, and I added an EMT-140 with an Abbey Road-like EQ curve in front of it (i.e. rolling off 600 Hz and 10 KHz). Finally, I added the UAD Precision Buss Compressor to level it out a bit. After exporting this, I pulled it into Wavelab, did a one-time gain uplift of 10 dB, and then used the Precision Limiter to raise it a few more dB to bring it up to standard listening levels.

 

The result is below:

 

March 06 Experiment

 

The noise floor is still really high, but when you think about it, it's not too surprising. Here's a patch that has been cut to smithereens, to the point where the full output is barely able to break -18 dBFS on a PPM meter. After more onboard EQ cuts, it's taking nearly 15 dB of boost to get it up to a listenable level. That's a lot of gain, whether it's applied one time or "live" with a gain/limiter plug-in.

 

Thanks also for cluing me in on the stereo issue. Some of these vintage reverb and delay plugins will sum the two input channels to mono, and I've never given that much thought. I will consider it moving forward.

 

I'm going to continue to work on this, and I may take my MP11 and layer a "real" Rhodes beneath the Roland it to give the sound a more healthy foundation. I can always roll-off the Kawai with a really steep LPF and use the Roland for the bell-like tones only. If I do this, I'll just kill off the boomy oscillator on the Roland.

 

If I can't get the right sound after all of this, I may just add a rain sample at low level throughout the track (to mask the noise issue). The title of the full song is "Condensation", since it reminds me of laying in bed on a weekend and staring at rain drops and mist on a window pane. There is a thunder sample at the end already, so a slight rain sound throughout wouldn't be out of place at all.

 

I'm going to write a new post about setting levels in a 24 bit environment. I've read several articles like this one, but I'm starting to wonder if I'm setting things too low. Since this is a solo EP piece, I don't have to worry about multiple tracks overloading the mix buss. But if my command of gain staging isn't right from the get go, nothing else can be right. I have to learn how to do this the correct way, so that I'm not constantly second guessing myself.

 

Thanks again for the help.

 

Todd

Link to comment
Share on other sites

  • Members

Takes work, this stuff, to get a handle on all the handles.

 

I'm thinking about the noise floor, having read your description. Your 3080 puts out a certain amount of noise that is just "there" in the electronics - it doesn't go up and down in level with the output level of your patch. So you want to start with a pretty strong signal coming out of the 3080 so the 3080 noise level is lower relative to the actual musical content it's putting out.

 

If you turn the 3080 patch way down, it's the same thing as turning UP the noise floor of the 3080. So I think you're increasing the noise content right out of your hardware rompler.

 

Then you're cutting this and that frequency in the DAW, and when you're done, you have to boost your track signal in your DAW - which floats up all the content in the signal, including the noise, giving it yet another boost. So your jacking up the noise floor in stages I think - at least if I'm understanding you correctly.

 

So, yeah, when you say you might need to bring up the signal coming out of the 3080, I think that's on the right track. Get that nice and hot. You can always turn down in the DAW without damaging the quality - it's boosting that usually brings the trash back in the house.

 

If I find myself with a cobbled-together DAW signal path that has me cutting, then boosting, then cutting again and/or even boosting yet again, I know I need to start over and create a simpler signal path - not one where I boost, which I have to later cut, then later boost, etc. If I need to cut 10db out of the midrange of a muddy VST instrument, I'll make sure that the level going into the EQ is such that I don't have to boost significantly again coming out of the EQ. So I just make one cut and that's pretty much it - no extra trips around the block. All the processing adds up the negatives that always come with processing.

 

About clipping from your midi rompler - what I do a lot with sampled instruments, is use Sonar's midi capabilities to restrict the velocity to some range, say 45-105 out of the 127 range. That's a way to compress through midi, so I don't have to tame huge peaks, or bring up deep troughs, in the level of a midi instrument. It's a bit tricky because velocity is not just a "volume" control - it's a sample-layer selecting mechanism also. So I'll tweak the restricted range to make sure I get the sounds I want, or avoid the sounds I don't want. But I never need a super-wide velocity range, so restricting the range does a lot to smooth the playback and make other processing easier and less invasive to the sound.

 

I hope all my input is helping and not wasting your time - some of the other brains at HC are better at this sort of thing than I am, although I'm right some percentage of the time :)

 

nat

 

 

 

 

Link to comment
Share on other sites

  • Members

I write a lot of music in a lot of styles and I find that the more synth sounds I use, the more I tend to edit the sounds at the source and then eq the sounds at the end. As someone already wrote, synths are usually so full that if you want them to get along and fit in the same space you really have to do that.

 

Cheers,

 

Mats N

Link to comment
Share on other sites

  • Members

About clipping from your midi rompler - what I do a lot with sampled instruments, is use Sonar's midi capabilities to restrict the velocity to some range, say 45-105 out of the 127 range. That's a way to compress through midi, so I don't have to tame huge peaks, or bring up deep troughs, in the level of a midi instrument.

 

Hi Nat. I'll occasionally do that as well in Cubase. I'll scale the velocity up or down as a whole, or if a few passages bother me, I'll edit them individually. Cubase also has a MIDI compressor plugin that allows you to do this in real time with a threshold control, etc.

 

 

If you turn the 3080 patch way down, it's the same thing as turning UP the noise floor of the 3080. So I think you're increasing the noise content right out of your hardware rompler.

 

When I added the onboard Stereo EQ in the XV-3080 and cut-out the boomy frequencies, that's when the output dropped significantly. If I run the original patch completely dry, there is plenty of signal. Once I apply the 400 Hz cut, the 630 Hz cut, the 800 Hz cut, and a gentle roll-off at 4 KHz, the gain is cut in half. Then when I record it in Cubase and do further dips with Fabfilter, that's when the noise floor really becomes a problem. The sound is cut so much, that raising the overall gain raises the noise floor. I think I've resolved that I'm going to add a rain sound in the background to mask it. Some tracks just aren't worth days upon days of effort, and rain will fit with the chords.

 

I hope all my input is helping and not wasting your time - some of the other brains at HC are better at this sort of thing than I am, although I'm right some percentage of the time :)

 

 

You've been a huge help. Thanks for taking the time to listen to the clips and provide advice. I'm continually learning, and in every major pursuit of my life, it's typically one or two generous individuals who provide a thread or insight that leads to the next big discovery, which then leads to the next one or two key individuals, which lead to the next discovery, and the cycle continues. It can be a fun ride.

 

Thanks again.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...