Need Advice on Plugin/Mix Settings and Setup

[Nokone Miku]

When I did my cover I processed the vocal track with: Equalizer > Digital Compressor > Tube Compressor > Analog Delay > Limiter. It took me a week to get the right plugins and tweak the settings.

The EQ was set to a "female pop vocal" preset that boosted the high-mids but had a high-high cut off at around 14k. I dropped a small, thin spike at the 1.2k mark to cut certain shrill peaks. I got interesting results if I set the gain to 'auto' but decided against it.

The digital compressor was set with an aggressive attack and release, which seemed to keep the high notes from getting too loud. One of the things that took me so long was figuring out where to set the input gain and output gain on this plugin. These settings probably had the biggest effect on the end result.

The tube compressor was fairly mild and I also used it to lower the overall volume to (I hope) match up the vocal volume with the volume of the music. I figured I should use the compressor plugin to do that, rather than just dropping the volume in the mixer. So the vocal track could be left at 0 db in the mixer. I ended up needing to set the instrumental track to -7 db but otherwise didn't alter it. This left the whole mixdown with a headroom of -2 db. I read that -2 db is ideal for uploading to Youtube and other streaming services, but I'm curious what you think.

The analog delay felt better than reverb for giving body to the vocal. I experimented between 10% and 14% wet mix and ended up around 12%. The limiter was set to -2 db and only triggered a few times throughout. I kept messing with everything until I felt the song sounded right on my good headphones, my crappy earbuds, my good speakers, and it sounded okay-ish on my cheap desktop speakers.

I'm also trying to figure out how much to adjust in the Piapro tuning and how much to leave to the EQ and compressor plugins. I can lower the volume and dynamics on problem notes in Piapro, but I don't know if that is the best way to handle it. (I really spent 10+ hours over the last month just on the line "the two of us" and the word "two" finally ended up being [ Sil T th U ] [ - u: w ] with a 3/32nd note gap between and something close to "fast 2" vibrato. Brightness up, dynamics down, and a small breathiness block under the "U." You can't even hear the "T" phoneme but it forces the "th" to be pronounced distinctly instead of being overpowered by the vowel sound. I could fill a page with all the phoneme and parameter variations I messed with. I kinda preferred [ Sil T th U ] [ - w ] with the "w" taking up half the note, but felt it still sounded a bit on the shrill side.)

Someday I'm going to do my own instrument track version in a slightly lower key and at 160 bpm instead of 180 bpm. I know I can get a more emotive performance that way.

Sorry this is a long write-up, but I want to make sure I have this stuff right before I finish my second cover. I also don't know if the single voice is sufficient or if I should have added a second vocal track to make it more substantial like the original version of the song has.

 

[inactive]

I got interesting results if I set the gain to 'auto' but decided against it.

I personally always EQ with auto gain enabled.

Ignore if you already know this, but when changes are made with an EQ, the volume of the signal also changes. Cuts make it quieter and boosts make it louder. Unfortunately, due to how human hearing works, we perceive an increase in volume as being better, even though that might not be the case. Auto gain compensates for this, allowing you to hear the post-EQ signal with the roughly the same volume level regardless of any boosts or cuts that have been made.

Of course, there are times when auto gain should be turned off, such as when you're deliberately pushing the high frequencies into a compressor when doing the Scheps vocal trick, but I leave it on because I get tricked by volume too easily.

I want to make sure I have this stuff right before I finish my second cover.

Unfortunately, there is no "right" way of doing this. As Chris Lord-Alge once said in an interview, mixing is an opinion.

 

[IO+]

That's a lot of questions, I'm not sure what are you aiming for.

It's nice if you give us your progression (audio files/SoundCloud) that you are working on, or maybe workflow/signal flow via picture so i get an idea.

Also i need to warn you that some plugins that have "auto gain" can be misleading, they work as normal gain increase not "gain compensation", so how to check it is quite simple. If you engaged the auto gain and it get instant louder, which mean the gain is not compensate.

 

[Nokone Miku]

That's a lot of questions, I'm not sure what are you aiming for.

I guess I'm asking if the mix on my cover sounds good? I don't have anyone around who's savvy about audio to ask. (They'll just tell me "it sounds fine.") I'm a bit paranoid that it only sounds right on my system. Sort of like when you do digital art and then figure out that the color temperature in you monitor's settings is too warm or the gamma is over-corrected and so your art looks different on everyone else's screen.

Is there anything obvious I'm missing or should try doing? Are there any "best practices" when it comes to processing Vocaloid tracks? Should I try and fix volume issues in Piapro or in the DAW? I kinda just need someone to talk about this stuff with. I don't know anyone who can give constructive criticism or suggestions on this sort of thing.

After tinkering around for weeks, this is what I ended up with on this vocal track (what I wrote in the first post was from memory). I tweaked things until the VU meters were bouncing with the music and not going below -10 db. I kept the delay mix low, because I'm afraid I'd sound like a noob if it was too obvious:



I know this is all subjective, but please tell me anything I ought to try, things to avoid. Or even just some things you've done that were cool or things that really went horrible. Or things that you thought were cool but then realized were horrible!

 

[IO+]

The EQ was set to a "female pop vocal" preset that boosted the high-mids but had a high-high cut off at around 14k.

Start with preset is nice, but i don't really use HPF on vocal, around 12-18kHz is where you get the "air" out of vocal, I rather boost with wide Q to get the most out of vocal that i can get.

The digital compressor was set with an aggressive attack and release, which seemed to keep the high notes from getting too loud.

I'm not sure what do you mean the high notes, but i assume that you mean "peaky waveform" aggressive peak are not common on organic vocal but synth vocal.
If you need to tame that "very fast aggressive" peak down. Please use limiter to do the job. Limiter are purposely design to eliminate the extremely fast transient in transparent way, due to the "look-ahead" algorithm, but compressor are opposite they're made to "control" not "eliminate".

The tube compressor was fairly mild and I also used it to lower the overall volume to (I hope) match up the vocal volume with the volume of the music. I figured I should use the compressor plugin to do that, rather than just dropping the volume in the mixer. So the vocal track could be left at 0 db in the mixer. I ended up needing to set the instrumental track to -7 db but otherwise didn't alter it. This left the whole mixdown with a headroom of -2 db. I read that -2 db is ideal for uploading to Youtube and other streaming services, but I'm curious what you think.

Doing gain compensate is a good habit. But fader are meant to be used so don't be afraid to use it.
About setting loudness for off vocal track. I usually listen the whole song a few time and measure average and short term loudness before i decided to low it down have enough room for vocals, it can be static or automated.

About loudness for youtube. That you heard is more like a myth (i might add this to my myth busting series.), but the only thing youtube care about is Loudness Units relative to Full Scale (LUFS) for youtube is -14 LUFS(short-term) How to check is quite simple. 1. go to the song that you upload on youtube 2. right click and select "stats for nerds" 3. looking for "volume / Normalized" 100% / ***% (content loudness ***dB) see how much you get penalty.

Example

This is my cover. I got 3.7dB normalized penalty.


Original song from Keeno got 3.7dB normalized penalty same as me.


My first cover back in 2016

As long as you not compress it so much and turn them into wreck cube. You perfectly fine.
I hope you get an idea.

The analog delay felt better than reverb for giving body to the vocal. I experimented between 10% and 14% wet mix and ended up around 12%.

I see you utilize the external bus which is a good thing for workflow, i suggest you try to utilize delay effect in separate fx channel or another bus channel, for flexibility. also you will have benefit of using fx on external bus which you can push the effect really hard and then adjust it with fx channel fader to taste.

I also don't know if the single voice is sufficient or if I should have added a second vocal track to make it more substantial like the original version of the song has.

it's depend on the song and what you want to achieve.

I guess I'm asking if the mix on my cover sounds good?

I can't legitimately answer your question without something that i can listen to, it's like i blindly give you advice/critique without knowing anything, it is a recipe for disaster.
I apologize if i sound paranoid or being too serious, i just can't do it. That's illegal for me.

I'm a bit paranoid that it only sounds right on my system.

You have good instinct. About reference you can read more in the resources thread. I think i made one in myth busting series.

Is there anything obvious I'm missing or should try doing? Are there any "best practices" when it comes to processing Vocaloid tracks?

I would say best practices is "Do thing with purpose, Not do thing for no reason"

Should I try and fix volume issues in Piapro or in the DAW?

If you mean vocal are too loud or too quiet, then fix the source, not in the mix.

I kinda just need someone to talk about this stuff with. I don't know anyone who can give constructive criticism or suggestions on this sort of thing.

I could give a help sometime, also you can check my profile for more information, but i might have a late response due to work.

 

[Nokone Miku]

I'm not sure what do you mean the high notes,

I mean high pitched music notes. For whatever reason Miku English gets really loud in volume on certain phonemes when she sings above G4.
Especially [ aI ], [ eI ], [ u: ], and [ i: ].

On my Youtube video the stats say "Volume/Normalized 42%/33% (content loudness 2.0 db)" what could I do to fix it in the future?

I've read all your audio myth busting series and looked at the plugins you posted. I'm not sure if I know enough yet to make full use of the plugins.

Thanks for your reply! It gives me some good things to think about.

 

[Nokone Miku]

In a lot of songs there are (I dunno what you call it) doubled-up vocals? Sometimes it's used to bulk-up a reedy or thin vocal. It usually isn't a chorus or echo plugin (I think?). I guess it's an actual second vocal track, either in the DAW or in the Vocaloid editor itself. Sometimes it sounds like they actually tuned a harmony track or some kind of accompaniment track.

Other times they just duplicated their main vocal track. Doing that usually causes distortion (would you call it interference?). I know there are things that you're supposed to do to a duplicated track to prevent that. I'm not sure if it has to do with sidechaining or equalizing or what.

And I'm not referring to backing tracks like the choral inserts in the song "Teo" by omoi.

Specifically right now I'm trying to figure out how to do something similar to the chorus of "Lost One's Weeping" by neru. (Not the bit preceding the chorus, but the chorus itself.)

I'm sitting here thinking to myself:

• Is that an echo or reverb or delay effect and I just don't know how to get it to sound right?
• Or just don't have the right type of plugin?
• Is it just pasting in a second vocal .wav file and then cranking the mids way down in the EQ?
• Or really heavily compressing it?
• Is it a second vocal track in the editor with the Brightness or Dynamics really low?
• Maybe it's also delayed by 1/64th note?
• Is it not a separate track and they're actually just splitting the main track between two buses with different effects?

I've tried looking it up, but I guess I don't know what terms to search for. When I search for "double vocal" it talks about having a singer do two takes and then panning them and stuff like that. Something I might mess with later, but not what I'm looking for right now.

It's probably something simple and I'm just doing it wrong. >_<



(EDIT): Reading more about it, maybe the doubling they're talking about is what I'm looking for? Something like a second duplicate track at lower volume with different EQ/compressor settings? Does that even work if the vocal tracks are exactly the same? (I just read an article that says they can't be exactly the same.)

I also saw a post talking about doing a thing where you offset the second track by -10 ms and then use a delay effect to push it back +10 ms with the feedback set to zero, and doing that gives the track a slightly different tone.

 

[IO+]

In a lot of songs there are (I dunno what you call it) doubled-up vocals? Sometimes it's used to bulk-up a reedy or thin vocal. It usually isn't a chorus or echo plugin (I think?). I guess it's an actual second vocal track, either in the DAW or in the Vocaloid editor itself. Sometimes it sounds like they actually tuned a harmony track or some kind of accompaniment track.

Other times they just duplicated their main vocal track. Doing that usually causes distortion (would you call it interference?). I know there are things that you're supposed to do to a duplicated track to prevent that. I'm not sure if it has to do with sidechaining or equalizing or what.

And I'm not referring to backing tracks like the choral inserts in the song "Teo" by omoi.

Specifically right now I'm trying to figure out how to do something similar to the chorus of "Lost One's Weeping" by neru. (Not the bit preceding the chorus, but the chorus itself.)

I'm sitting here thinking to myself:

• Is that an echo or reverb or delay effect and I just don't know how to get it to sound right?
• Or just don't have the right type of plugin?
• Is it just pasting in a second vocal .wav file and then cranking the mids way down in the EQ?
• Or really heavily compressing it?
• Is it a second vocal track in the editor with the Brightness or Dynamics really low?
• Maybe it's also delayed by 1/64th note?
• Is it not a separate track and they're actually just splitting the main track between two buses with different effects?

I've tried looking it up, but I guess I don't know what terms to search for. When I search for "double vocal" it talks about having a singer do two takes and then panning them and stuff like that. Something I might mess with later, but not what I'm looking for right now.

It's probably something simple and I'm just doing it wrong. >_<



(EDIT): Reading more about it, maybe the doubling they're talking about is what I'm looking for? Something like a second duplicate track at lower volume with different EQ/compressor settings? Does that even work if the vocal tracks are exactly the same? (I just read an article that says they can't be exactly the same.)

I also saw a post talking about doing a thing where you offset the second track by -10 ms and then use a delay effect to push it back +10 ms with the feedback set to zero, and doing that gives the track a slightly different tone.

Im not sure what do you want to achieve or what do you're trying to do, maybe you can give me more example or TL:DR version.

Chorus is actually not so complicate as you thought, it just a simple chord, since main key of vocal is already monophonic, if you have some basic keys knowledge, you're ready to go!

 

[Nokone Miku]

Sorry for my rambling stream-of-consciousness post.

The thing I'm trying to do here-and-now is approximate the sound of "Lost One's Weeping" by Neru.

For the effect from 1:02 to 1:12 I can get something I like from either an analog delay plugin or a tube distortion effect plugin. Something like that.

But I'm not sure exactly what to do for the chorus (1:14 to 1:24) and (1:26 to 1:36).

Since my post I've been experimenting with the vocal doubling techniques I've been reading about. I'm thinking now that the chorus isn't using an effect. It seems like they used two separate vocal tracks for those parts and overlaid them.

I don't think the second track is actually tuned as a harmony. It sounds like it's a duplicate track with a different compressor on it.

 

[inactive]

The sections you pointed to sound like two voices singing in (mostly) thirds.

And, wow, that track over-compressed and too loud. Out of curiosity, I ripped the audio and measured true peaks at 2.0 dB above zero. Geez...

 

[IO+]

1:02 to 1:12 sound like they using different vbs or using simple lpf
for the chorus (1:14 to 1:24) and (1:26 to 1:36) I think it's a simple chord, a full 1oct lower and 1oct higher.

I try to replicate them as close as i can, and i'm sorry for the poor voice programming.
This is the result.

(my soundcloud is running out of space, i will delete it soon.)

This is the picture of chord that i use in vocaloid, i've put it on midi editor so you can see all the progression.
The whole chorus look like this and repeat for the entire song, there is some minor key on very small part, but that's about it.


I hope this help.

 

[Nokone Miku]

So, I'm using the Redlight Distortion plugin that comes with Studio One. For some reason it takes about two seconds to kick in, so the first two words of the line of lyrics don't have the distortion applied to them. (Maybe it is because the delay effect needs some sound to delay, so it has to wait until after there is sound already playing? I guess?)

The only workaround I can think of is to add a couple extra words before that section, send the section to a bus, and then automate the bus to mute its volume while those extra words play, then unmute the output where the actual lyrics begin.

Or, put in the extra words and have the section as a separate audio track. Render it out with all the relevant plugins and such. Then import it and trim off the extra words.

I doubt there's a way to make a plugin 'look ahead' if the feature isn't a setting on the plugin itself.

 

[IO+]

Sound like something goes wrong, did your system running smoothly?

Or maybe you simply create automation by accident, try to check the automate that exists.

 

[Nokone Miku]

The automation is working correctly. I thought it might be my system overloading/lagging, but, after testing several different things, that isn't the case. Raising the input gain helped a little? But it still seems like it is taking a moment for the plugin effect to kick in, or become noticeable. I dunno, I'll mess around with it some more later.

On a different topic:

Vocal Resonance

I used to think the polyphonic distortion (what I called it) I heard in Vocaloid songs was a fault of the Vocaloid software. Now I realize that it is the result of people doubling their vocal track and then not dealing with the resultant resonance that it creates. (I was looking up terms like interference, dissonance, etc until I figured out it was called resonance.)

Out of around a dozen articles I read, these two seemed to be the best:

EQing Resonant Frequencies and Harsh Sounds​

How To Use A Vocal EQ Chart​

I thought I'd ask if anyone has additional advice or advice specific to Vocaloid?

 

[Nokone Miku]

[regarding: Methods to make a metal scream/death growl with VOCALOID]

Seems that the Charsiesis and Vocascreamer plugins aren't compatible with my version of 64-bit Studio One. And I don't really want to run them in a separate program.

It's so hard finding plugins relevant to voice synth/Vocaloid. So many "vocals" plugins are oriented around cleaning up and mixing imperfect voice recordings. Autotuning, microphone noise removal, adding saturation or warmth or various flavors of reverb. It's mostly stuff that can be achieved with the Vocaloid parameters, or a basic compressor. I downloaded one yesterday and when I was cranking the virtual dial on it I thought, "This really sounds no different than raising or lowering the Brightness parameter."

I feel like so many other plugins are just oriented around messing with guitar sounds or synth samples. Now granted I'm only looking at free VST plugins. At this point I don't know what to look for in a commercial plugin. I don't want to drop $100 on something and then find out I either don't know how to use it fully or it doesn't do what I want.

Part of the problem, I suppose, is that I'm trying to do realistic vocals. So a lot of distortion plugins that might be useful to others just make it sound too synthetic for what I'm doing. Sorry if I just sound like I'm griping.

Anyway, I've been seeing what I can do within Piapro. I've gotten some promising results using the Growl parameter, Vibrato, and scribbling in the Pitch Bend with the quantization turned off. I'm not trying to get a full on metal scream, I just need vocal fry and some "shouting" inflection to add emphasis to certain words. I'm trying to think of examples, like, maybe a much lighter version of what Utsu-P did with "Not-Photogenic." Eh, not quite like that. It's more something you get with female punk vocalists, but I'm having trouble finding a good example.

Honestly, I'm spending a stupid amount of time on just five syllables in the song. But I feel like those five syllables are going to make or break to whole thing.

 

[Nokone Miku]

It took all weekend, but I finally came up with what I would call a "harsh shout" using only parameters. (At least it sounds good to me, I'm not sure how it will sound to others.) I'm sure part of my struggle is getting Miku to sound forceful at higher pitches. The big problem is that very minor adjustments can screw it all up. To put it vaguely it involved:

• brightness peaking during vowels sustained in the middle of syllables
• dynamics dropping sharply to nothing between each word
• growl rising in the latter half of a word
• pitch jumping a tiny bit up in the first 1/64th or 1/32nd and then dropping a smidge flat before trailing off at the very end
• clearness spikes on certain consonants
• gender adjustment on words that need it
• vibrato had to be applied hard just before half way through the word.
• breath applied to the ends of higher pitched notes
• and often a 1/64th [- Sil] between words

And the final breakthrough that gave it the last little thing it needed was using pitch bend in just the right places making a tiny sawtooth followed by a reversed sawtooth, like:
/ / / /\ \ \ \
For some reason I started calling it a "crown of thorns." Sort of how I think of other curves as a "shark fin" or "ski slope" or "fangs."

So... uh, like I said, it took all weekend.

 

[Nokone Miku]

Reading this article on Gain Staging was a revelation for me. I've been cranking everything up to almost full volume right at the start of the mix chain. No wonder the plugins were so touchy, and tiny, little adjustments were having huge effects.

What is Gain Staging?
In general terms, gain staging is about adjusting the volume at each point of amplification a signal passes through on its way to the master buss. “Points of amplification” can include preamps, channel faders, plugins or hardware signal processing — basically anywhere you have the option to adjust volume as signal passes through. The goal is to set levels at each point of amplification so that the following point receives an ideal level of signal.

So, me having the volume at full blast going into each plugin along the chain has been limiting their usefulness and range of effects.

(cue the Ive_been_so_stupid.gif)

 

[Nokone Miku]

After learning this and then dwelling on it for an hour, and thinking about all I had learned... I had to go outside and scream at the sky:

"This is it? This was the problem!? This is why I wasted entire evenings wondering why it didn't sound right and the plugins weren't working how I thought they should!!? The goddamn Gain was set too high!!!? AAAAAAAaaaaaahhhhh!"

 

[Nokone Miku]

Reworking all of my mixes and plugins and stuff. Things make a lot more sense now.