Moderators: MattKingUSA, khz
Hello everybody,
I started tinkering with equalizers and compressors to improve (hopefully) my recording of piano solo (jazz, SFZ ibraries mainly).
I watched a few tutorials, but most deal with mixing the piano with other instruments, and other stuff I'm not gonna need.
My immediate goals are to soften the brightness of the mid range and to improve the balance between soft and loud notes.
So, if anyone has any tip or suggestion they are very welcome.
I use Auburn Sounds Lens for things like that. It's a multiband (up to 64 bands!) compressor with sidechain EQ to tailor the compression for each frequency or range of frequencies. It will make your head spin trying to understand it at first but well worth learning.
There's even a free native Linux version! 
https://www.auburnsounds.com/products/Lens.html
Thanks for the suggestion.
I downloaded Lens and will try it out.
I have limited needs, basically just record videos in OBS Studio for my teacher and friends.
So far, I used the LSP plugins and also VLC inbed EQ and compressor.
More than the software, I'm interested in which parameters I should modify and how in order to improve the piano sound.
I'm aware that this is a very general question, but any tip is welcome.
For EQ I always look at the 200-500hz range since it can get really muddy if there is too much in there. However, if you want to soften/darken the piano I would try a high shelf, or if you want to keep some of the higher end and want to remove some of the high mids then use a wide Q bell shape.
Compressors are really tricky to recommend settings but you would want to play with attack, release, threshold, ratio. I'd probably start by bringing the threshold down more than you would want so you can hear it working, raise the ratio fairly high (like 4:1 maybe), then play with the attack and release, maybe 20-30ms attack and a longer release, maybe 100-150ms. Once it is behaving how you want, raise the threshold until it is only affecting what it needs to, and then decrease the ratio so it's not so aggressive. Really depends what you want to get out of it though and many other factors. There are many approaches too, you could go for high ratio and high threshold (often useful for dealing with peaks) or low ratio and low threshold (useful for levelling). You can also do parallel compression so you mix a fully wet and usually heavily compressed signal in with the original dry signal, this will raise the quiet parts but keep the dynamics of the overall sound. You could also look at different types of compression, RMS might be useful for you.
Whatever you do with the compressor, set the makeup gain so that when you disable the compressor the level is the same as it was before, this way you are not tricking your ear by mere volume and can accurately judge whether the effect is doing what you want it to by bypassing it and enabling it again. I like to close my eyes when doing this and hit bypass a bunch of times so I don't know when it is enabled, and then choose which sounds better, if it's not the enabled version then you need to change things (or not use it at all).
Very decent recommendations so far.
I'd like to add some considerations regarding the signal/work-flow.
Think in stages. You don't wanna tackle your compressor with unwanted parts of the source signal.
Gain automation might help to reign in parts where notes poke out
This can be of much greater use in ensemble mixes where the level difference of one instrument has a bigger range of not being audible. But even on a solo instrument this can be used without sounding unnatural.
Substractive eq
High-pass filtering may reduce unwanted rumble or (with a gentle curve) soften the low-end alltogether.
As @sysrqer pointed out, the low-mid area is more often than not a bit convoluted with resonances. That is a much bigger issue with mic recordings of real instruments but not entirely exclusive to those. I often start with a dip in the wolvetone area (500Hz) and see/hear if the tone of the instrument "opens up".
Resonances and harsh sounding overtones in the higher mid area are often much harder to tackle. You need much narrower dips and they usually only appear every now and then. One can find them with a parametric eq and either automate the dips or use the found frequency settings and feed them in one or more instances of Zam DynamicEQ and adjust the threshold, attack and release settings. This is a bit fiddly and maybe only workable if there are severe problems to solve.
Compression
Most settings are really dependent on the source material you feed into it. It might be a good idea to use a compressor with sidechain filter capabilities to not let the impressive bass line pull everything else down.
EQ Boosts
Finetune
If you have the feeling that your 5k boost for that extra sparkle could even go higher but it also pulls up some unwanted harshness, you could try light de-essing, tape saturation, and/or multiband or dynamic compression another compressor with a very high low-pass sidechain filter setting to only compress the high-mid region (Fircomp2 has a really workable preset "Bus De-Harsh" for that).
With Lens you can do most of 3-5 and some bits of 2. For me it is still kind of black-belt level art, though.
Thanks a lot guys,
You gave me quite a bit to work with. I need to educate my era to perceive the differences.
Something I didn't mention is that I'd rather modify the recorded track rather than work on the live sound. Don't know if that makes any difference as for your suggestions, but felt it was worth mentioning.
I'm gonna record a test track and try to apply your tips.
Yes definitely record it first then modify it, lest you remove things on the way in which you later decide you shouldn't have.
The reason I prefer to use multiband compression or dynamic EQ to straight EQ is that instruments such as piano change timbre depending on their level. If you tinkle the ivories softly you get a sweet sound, but if you hit the keys hard you may introduce some annoying harshness. Dynamic processing can be set to react minimally or not at all for the softer sounds but more for the louder ones.
It can be a dark ninja art though. Just leaving autoplay rolling on Soundcloud will bring up many instances of what not to do! 
So very true.
Even on a digital, the way you play each note has a major effect on the timbre of the sound.
I wish I could equalize myself into Maria Joao Pires, but it's sadly not in the cards.
Since one of the libraries I play is the Salamander, on which Orchestools Piano is based on, I studied the OT presets closer to the sound I want.
Here are the EQ curves:
The Mellow preset is close enough, although the trebles are a somewhat unnatural, but at least I have some reference points..
You could take the EQ curve from one of the recordings and apply it to your recording although I'm not aware of any linux plugins to do this (it is a thing though).
What is OT? I'm not sure what we're looking at here.
What is OT? I'm not sure what we're looking at here.
Orchestools.
Its piano plugin is based on the Salamander library, and comes with several presets.
Each preset has a different EQ setting (among other things I guess, including compression). These are the curves of the 3 presets I like best.
The Salamander is one of the 2/3 piano libraries I use. Although Orchestools basic piano doesn't sound exactly like the Salamander, it's still close enough.
In fact, it still retains the slight metallic twang in the octave across middle C that I'm trying to get rid of.
I figured these EQ curves might give me a hint of how a certain piano sounds when modified differently.
I haven't yet had the chance to really tinker with them, just recorded the same piece a few times with each and tried to put my finger on what I like and what I don't. A starting point to educate my ear to identify the fine differences beyond the overall impression.
There are several ways to do this in Linux:
I think they all work in similar ways, by creating an impulse response of the EQ of one file and applying it to another file. MelMatchEQ may be the easiest to use as it is the only one with a GUI.
I tried them all recently trying to breathe some life into some old flat-sounding vinyl recordings. In the end however I got the best results from Lens.
sunrat wrote: ↑Mon Aug 28, 2023 7:41 amThere are several ways to do this in Linux:
- Timbre-mill from@sadko4u - there's a topic about it here - viewtopic.php?p=129546
- MelMatchEQ from Hermann Brummer - https://github.com/brummer10/MelMatchEQ.lv2
- eq_match in the fps plugins - viewtopic.php?p=156608#p156608
I think they all work in similar ways, by creating an impulse response of the EQ of one file and applying it to another file. MelMatchEQ may be the easiest to use as it is the only one with a GUI.
I tried them all recently trying to breathe some life into some old flat-sounding vinyl recordings. In the end however I got the best results from Lens.
That was actually a joke about EQ myself into playing like Maria Joao.
But your suggestions are very interesting.
I got that you were joking. I'd like to EQ myself into Larry Carlton. 