Processing Vocals - Parallel vs. Plain Compression
Vocals are one of the most dynamic sounds we can record, and there can be a lot of dynamic changes, hence we barely can get away without compression.
One method is a general compressor, set up with a tasteful amount of attack that even adds a bit of punch (though I want to keep sounds as natural as possible). Problem with this is, that it not only controls the macro dynamics (big changes), but also squashes the more decent transients a bit. Not the best sound, let's face it.
Let's not worry about the remaining peaks - you do either a serial, secondary compression/limiting (in my preference, multi-band soft clipping), or whatever you prefer. De-essing frequency limiting is also another topic.
Back to the raw recording, there's the other solution of parallel compression. Some do light compression, or none on the raw vocals, and a harder, parallel processing. It's a fact, that this one blends the original dynamics, but the parallel processing's god awful sound will still be audible. It will be there, no matter what. The special thing, though is, that this method still adds some wanted transients, that the classical compression affects rather badly.
This results in some awful agony. Dx Average dynamic destruction with a clean sound, or parallel processing with added dirt?
Don't even say "why not parallel process the classically compressed sound with the raw recording", because it will be insufficient. I'd end up dialing an almost completely wet signal anyway.
The only solution seems a trade-off with the dirt at the moment. Aka - creating both processing methods, and then make a parallel mixing between the two like so: put them into a crossfader, then fade to taste (just like dry/wet, only classical/modern).
Welp. Thoughts, ideas?
One method is a general compressor, set up with a tasteful amount of attack that even adds a bit of punch (though I want to keep sounds as natural as possible). Problem with this is, that it not only controls the macro dynamics (big changes), but also squashes the more decent transients a bit. Not the best sound, let's face it.
Let's not worry about the remaining peaks - you do either a serial, secondary compression/limiting (in my preference, multi-band soft clipping), or whatever you prefer. De-essing frequency limiting is also another topic.
Back to the raw recording, there's the other solution of parallel compression. Some do light compression, or none on the raw vocals, and a harder, parallel processing. It's a fact, that this one blends the original dynamics, but the parallel processing's god awful sound will still be audible. It will be there, no matter what. The special thing, though is, that this method still adds some wanted transients, that the classical compression affects rather badly.
This results in some awful agony. Dx Average dynamic destruction with a clean sound, or parallel processing with added dirt?
Don't even say "why not parallel process the classically compressed sound with the raw recording", because it will be insufficient. I'd end up dialing an almost completely wet signal anyway.
The only solution seems a trade-off with the dirt at the moment. Aka - creating both processing methods, and then make a parallel mixing between the two like so: put them into a crossfader, then fade to taste (just like dry/wet, only classical/modern).
Welp. Thoughts, ideas?
Are there any examples of vocals you've ever heard that don't suffer from the problems you describe above? If so, then there are ways to achieve what you want without any trade offs.RobC wrote: ↑06 Apr 2018Vocals are one of the most dynamic sounds we can record, and there can be a lot of dynamic changes, hence we barely can get away without compression.
One method is a general compressor, set up with a tasteful amount of attack that even adds a bit of punch (though I want to keep sounds as natural as possible). Problem with this is, that it not only controls the macro dynamics (big changes), but also squashes the more decent transients a bit. Not the best sound, let's face it.
Let's not worry about the remaining peaks - you do either a serial, secondary compression/limiting (in my preference, multi-band soft clipping), or whatever you prefer. De-essing frequency limiting is also another topic.
Back to the raw recording, there's the other solution of parallel compression. Some do light compression, or none on the raw vocals, and a harder, parallel processing. It's a fact, that this one blends the original dynamics, but the parallel processing's god awful sound will still be audible. It will be there, no matter what. The special thing, though is, that this method still adds some wanted transients, that the classical compression affects rather badly.
This results in some awful agony. Dx Average dynamic destruction with a clean sound, or parallel processing with added dirt?
Don't even say "why not parallel process the classically compressed sound with the raw recording", because it will be insufficient. I'd end up dialing an almost completely wet signal anyway.
The only solution seems a trade-off with the dirt at the moment. Aka - creating both processing methods, and then make a parallel mixing between the two like so: put them into a crossfader, then fade to taste (just like dry/wet, only classical/modern).
Welp. Thoughts, ideas?
Audio examples can take the place of a wall of text when you're asking questions like these…
You may want to look into some old school solutions, aka "riding the fader". Good engineers often do this when recording very dynamic sources, sometimes before they hit any compression at all, as they are tracking the vocals. If you're not confidant doing that, then record the vocal with no compression, and flatten out any extreme dynamics BEFORE hitting any compression. This gives the compression less work to do!
Some times you have to "hand craft" your sound, and not rely on the tools to do all the heavy lifting for you, or you may end up with unwanted artifacts.
Selig Audio, LLC
The key is knowing why you're using compression. Don't compress just to compress. Have and know the reason why you're doing it. If you're using mild light compression on just the highest peaks it shouldn't affect smaller dynamics nor affect transients enough to sound "bad". That would mean you set the threshold too low and/or ratio too high. Also remember compressors can either add punch or subtract it depending of the attack and release times. If you set the settings right (dependent on material), the compression will sound natural. Parallel compression is about bringing up the volume of lower dynamics of the vocal without crushing the main vocal in the mix, hence the heavy compression because the louder dynamics of the vocal aren't needed in the parallel channel.
- Timmy Crowne
- Competition Winner
- Posts: 357
- Joined: 06 Apr 2017
- Location: California, United States
This.
And this.
Mixing is like practicing medicine. The least invasive procedure is the best, and avoid unnecessary squeezin'.
You know, I never really gave others' vocals that much thought, but if we pay closer attention, it could be that perfect recordings aren't exactly common.selig wrote: ↑06 Apr 2018Are there any examples of vocals you've ever heard that don't suffer from the problems you describe above? If so, then there are ways to achieve what you want without any trade offs.
Audio examples can take the place of a wall of text when you're asking questions like these…
You may want to look into some old school solutions, aka "riding the fader". Good engineers often do this when recording very dynamic sources, sometimes before they hit any compression at all, as they are tracking the vocals. If you're not confidant doing that, then record the vocal with no compression, and flatten out any extreme dynamics BEFORE hitting any compression. This gives the compression less work to do!
Some times you have to "hand craft" your sound, and not rely on the tools to do all the heavy lifting for you, or you may end up with unwanted artifacts.
Now when it comes to my own vocals, of course I will hear even what's not there at some point. xD
True, but it was quicker to type out than creating examples this time.
Don't worry, I do consider those macro dynamics, hell even while recording, behind the microphone (watching distance, and performance loudness). And of course I do some pre-mixing, so everything is more or less the same audio level, but some compression is still inevitable - and I want that to sound as transparent as possible. Otherwise it gets lost in the song mix.
I compress, so I don't have to manually balance the loudness of smaller differences out. Thing is, I do look for some added punch, but yeah, that will affect the natural sound, so I'd have to do some parallel mixing with the raw sound - but that might not be balanced enough then. Thus I'd need to make a secondary group of harder parallel processing, and then mix it with the softly processed group.QVprod wrote: ↑06 Apr 2018The key is knowing why you're using compression. Don't compress just to compress. Have and know the reason why you're doing it. If you're using mild light compression on just the highest peaks it shouldn't affect smaller dynamics nor affect transients enough to sound "bad". That would mean you set the threshold too low and/or ratio too high. Also remember compressors can either add punch or subtract it depending of the attack and release times. If you set the settings right (dependent on material), the compression will sound natural. Parallel compression is about bringing up the volume of lower dynamics of the vocal without crushing the main vocal in the mix, hence the heavy compression because the louder dynamics of the vocal aren't needed in the parallel channel.
But I've also thought of doing a group processing where A is a very subtle balancing compression, and B is the punchy one as the "hard" parallel processing.
If confusing:
(X meaning crossfading, like Dry/Wet)
A = Raw X Subtle
B = Raw X Punchy
Then
A X B
Of course, I do agree!WeaponX323 wrote: ↑06 Apr 2018This.
And this.
Mixing is like practicing medicine. The least invasive procedure is the best, and avoid unnecessary squeezin'.
But also that: "I want to keep sounds as natural as possible" - doesn't that already say it?
Imho, you're over complicating it. If you want added punch then you don't want the raw sound. What you describe seems to be changing the rawl sound simply to mix it back in giving you the same "problems" you started with. If you want a natural vocal sound then you can just do light compression (about medium attack and mid to slow release) or even a small amount of limiting on the vocal and it will still sound natural. 3dB of gain reduction (sometimes even less) is plenty. If you need more compression than that to level out small differences then you need to manually adjust your audio clips.RobC wrote: ↑06 Apr 2018I compress, so I don't have to manually balance the loudness of smaller differences out. Thing is, I do look for some added punch, but yeah, that will affect the natural sound, so I'd have to do some parallel mixing with the raw sound - but that might not be balanced enough then. Thus I'd need to make a secondary group of harder parallel processing, and then mix it with the softly processed group.QVprod wrote: ↑06 Apr 2018The key is knowing why you're using compression. Don't compress just to compress. Have and know the reason why you're doing it. If you're using mild light compression on just the highest peaks it shouldn't affect smaller dynamics nor affect transients enough to sound "bad". That would mean you set the threshold too low and/or ratio too high. Also remember compressors can either add punch or subtract it depending of the attack and release times. If you set the settings right (dependent on material), the compression will sound natural. Parallel compression is about bringing up the volume of lower dynamics of the vocal without crushing the main vocal in the mix, hence the heavy compression because the louder dynamics of the vocal aren't needed in the parallel channel.
But I've also thought of doing a group processing where A is a very subtle balancing compression, and B is the punchy one as the "hard" parallel processing.
If confusing:
(X meaning crossfading, like Dry/Wet)
A = Raw X Subtle
B = Raw X Punchy
Then
A X B
If you still say need more punch after that, then sure parallel compression can be an option for that, but you're supposed to mix the parallel signal in to enhance the natural sounding vocal. Not vice versa (yes it matters). Also its not necessarily the same as cross fading. Cross fading decreases one signal as you increase the other. Sure that's how a dry/wet knob works, but when dealing with faders on a mixer, the more natural vocal is at a static value while you mix in the parallel to enhance it.
It could also be, that I need to give examples after all, cause I tend to sound confusing. xDQVprod wrote: ↑06 Apr 2018Imho, you're over complicating it. If you want added punch then you don't want the raw sound. What you describe seems to be changing the rawl sound simply to mix it back in giving you the same "problems" you started with. If you want a natural vocal sound then you can just do light compression (about medium attack and mid to slow release) or even a small amount of limiting on the vocal and it will still sound natural. 3dB of gain reduction (sometimes even less) is plenty. If you need more compression than that to level out small differences then you need to manually adjust your audio clips.
If you still say need more punch after that, then sure parallel compression can be an option for that, but you're supposed to mix the parallel signal in to enhance the natural sounding vocal. Not vice versa (yes it matters). Also its not necessarily the same as cross fading. Cross fading decreases one signal as you increase the other. Sure that's how a dry/wet knob works, but when dealing with faders on a mixer, the more natural vocal is at a static value while you mix in the parallel to enhance it.
Thing is, during sound design, I usually tried setting both sounds to approximately even levels, and then crossfade, cause that way I could hear how much changes are happening to the sound. Mixing to a static level resulted in adding incorrect amounts for me, when it came to (originally) a single sound..
If you mixing in a parallel signal and you can't really hear the change then the problem is probably the processing on the parallel. Parallel compression means your compressing at extreme settings. If the settings are correct for the application you want, it should be very noticeable when you add too much of the highly compressed parallel signal to the static signal.RobC wrote: ↑06 Apr 2018It could also be, that I need to give examples after all, cause I tend to sound confusing. xDQVprod wrote: ↑06 Apr 2018Imho, you're over complicating it. If you want added punch then you don't want the raw sound. What you describe seems to be changing the rawl sound simply to mix it back in giving you the same "problems" you started with. If you want a natural vocal sound then you can just do light compression (about medium attack and mid to slow release) or even a small amount of limiting on the vocal and it will still sound natural. 3dB of gain reduction (sometimes even less) is plenty. If you need more compression than that to level out small differences then you need to manually adjust your audio clips.
If you still say need more punch after that, then sure parallel compression can be an option for that, but you're supposed to mix the parallel signal in to enhance the natural sounding vocal. Not vice versa (yes it matters). Also its not necessarily the same as cross fading. Cross fading decreases one signal as you increase the other. Sure that's how a dry/wet knob works, but when dealing with faders on a mixer, the more natural vocal is at a static value while you mix in the parallel to enhance it.
Thing is, during sound design, I usually tried setting both sounds to approximately even levels, and then crossfade, cause that way I could hear how much changes are happening to the sound. Mixing to a static level resulted in adding incorrect amounts for me, when it came to (originally) a single sound..
- Timmy Crowne
- Competition Winner
- Posts: 357
- Joined: 06 Apr 2017
- Location: California, United States
I hear you, and you’re right that we rarely if ever hear perfect, yet uncompressed, production-quality vocals. We don’t always notice how unnatural a vocal is because we rarely have access to the original recording.
With parallel tracks, it’s also good to color it with some EQ or filtering because the aggresive compression usually alters the tone in an unintended way. Too quick attack and release settings can cause chattering or distortion.
OK, that genuinely made me laugh out loud, in addition to being 'sound' advice.WeaponX323 wrote: ↑06 Apr 2018This.
And this.
Mixing is like practicing medicine. The least invasive procedure is the best, and avoid unnecessary squeezin'.
Selig Audio, LLC
OK, that's problem #1 - you're possibly trying to find a short cut to something where the long way is the way that will give you what you ACTUALLY want. Almost all compressors give a "sound", which I (and others) happen to LOVE!
BUT, if you don't love that sound (and it's not 100% right for every project IMO), you need to learn to ride the fader. This takes practice, but I've yet to find an "automatic" tool that can give me what I want to hear (because we all want to hear something slightly different, and it varies from song to song, from singer to singer).
Yes, definitely over thinking it!RobC wrote: ↑06 Apr 2018Thing is, I do look for some added punch, but yeah, that will affect the natural sound, so I'd have to do some parallel mixing with the raw sound - but that might not be balanced enough then. Thus I'd need to make a secondary group of harder parallel processing, and then mix it with the softly processed group.
But I've also thought of doing a group processing where A is a very subtle balancing compression, and B is the punchy one as the "hard" parallel processing.
If confusing:
(X meaning crossfading, like Dry/Wet)
A = Raw X Subtle
B = Raw X Punchy
Then
A X B
You also need to consider the source. This starts with the performer, the space they are in, the microphone, and the rest of the chain and how you set it all up. Those are some of the biggest variables right there, most of which cannot be overcome by some clever setup of compressors, EQs, and crossfaders.
Even with the same singer on the same setup, I need totally different approaches to vocal levels. On some songs the automation looks like a NYC skyline, with different levels for each word, even each syllable. Then on the very next song, I may not need ANY fader rides at all! There is simply not a "one size fits all" for every song, and you have to learn to listen and decide what's actually needed on a song by song (sometimes on a section by section) process.
Like QV said, "Don't compress just to compress". That right there is GOLDEN, I tells ya!
Selig Audio, LLC
I did hear the changes, but I either added too much, or too little. It's like the case of distortion. A sound won't necessarily become "harder", louder, if we just clip it - but we will really hear the effect if with the same amount it's clipped, we reduce its audio level at the same time.QVprod wrote: ↑06 Apr 2018If you mixing in a parallel signal and you can't really hear the change then the problem is probably the processing on the parallel. Parallel compression means your compressing at extreme settings. If the settings are correct for the application you want, it should be very noticeable when you add too much of the highly compressed parallel signal to the static signal.RobC wrote: ↑06 Apr 2018
It could also be, that I need to give examples after all, cause I tend to sound confusing. xD
Thing is, during sound design, I usually tried setting both sounds to approximately even levels, and then crossfade, cause that way I could hear how much changes are happening to the sound. Mixing to a static level resulted in adding incorrect amounts for me, when it came to (originally) a single sound..
In other words, the added audio level kind of fools my ears. At least that's how I see it.
I must say, I even heard pretty bad vocals on older vinyls, yet they knew their $h!t...WeaponX323 wrote: ↑06 Apr 2018I hear you, and you’re right that we rarely if ever hear perfect, yet uncompressed, production-quality vocals. We don’t always notice how unnatural a vocal is because we rarely have access to the original recording.
With parallel tracks, it’s also good to color it with some EQ or filtering because the aggresive compression usually alters the tone in an unintended way. Too quick attack and release settings can cause chattering or distortion.
Yep, that's why I rather wanted to say parallel processing. Either way, once the compression and de-essing is done, I would start my specific frequency spectrum balancing equalization, etc.
About #1 I'm okay with a trade off, but there's only so much we can manually do, BUT I'll still give it a try and compare things.selig wrote: ↑06 Apr 2018OK, that's problem #1 - you're possibly trying to find a short cut to something where the long way is the way that will give you what you ACTUALLY want. Almost all compressors give a "sound", which I (and others) happen to LOVE!
BUT, if you don't love that sound (and it's not 100% right for every project IMO), you need to learn to ride the fader. This takes practice, but I've yet to find an "automatic" tool that can give me what I want to hear (because we all want to hear something slightly different, and it varies from song to song, from singer to singer).
Yes, definitely over thinking it!RobC wrote: ↑06 Apr 2018Thing is, I do look for some added punch, but yeah, that will affect the natural sound, so I'd have to do some parallel mixing with the raw sound - but that might not be balanced enough then. Thus I'd need to make a secondary group of harder parallel processing, and then mix it with the softly processed group.
But I've also thought of doing a group processing where A is a very subtle balancing compression, and B is the punchy one as the "hard" parallel processing.
If confusing:
(X meaning crossfading, like Dry/Wet)
A = Raw X Subtle
B = Raw X Punchy
Then
A X B
You also need to consider the source. This starts with the performer, the space they are in, the microphone, and the rest of the chain and how you set it all up. Those are some of the biggest variables right there, most of which cannot be overcome by some clever setup of compressors, EQs, and crossfaders.
Even with the same singer on the same setup, I need totally different approaches to vocal levels. On some songs the automation looks like a NYC skyline, with different levels for each word, even each syllable. Then on the very next song, I may not need ANY fader rides at all! There is simply not a "one size fits all" for every song, and you have to learn to listen and decide what's actually needed on a song by song (sometimes on a section by section) process.
Like QV said, "Don't compress just to compress". That right there is GOLDEN, I tells ya!
When I'm the performer myself though, then I have the true physical control, so I hear how I need to move; and interestingly I once managed to perform in a "compressed, de-essed" way. Problem was that afterwards, I spoke like that, too for a short time. xD
Okay, that fader-riding as you describe it, sounds like a hell lot of work. I feel the automatic vocal slicing for REX - suggestion I posted the other week - would come in handy.
Now, I only compress when needed. -> As for the crossfader thing - it can happen that I find myself turning it totally dry.
I have to find an efficient solution from these with some trade-offs.
Thank you everyone, again!
- Marco Raaphorst
- Posts: 2504
- Joined: 22 Jan 2015
- Location: The Hague, The Netherlands
- Contact:
McDSP Moo and Klanghelm MJUC (VST/AU) can do super slow compression which won't sound like a compressor but is more like riding the vocal with a fader. This sometimes works like magic for example on very complex voices with a lot of dynamics.
I also use iZotope Declick a lot because I hate the "mouth drop" noises a voice often makes, and most compressor can crazy on these things. I kill these out of the signal most of the time. Hate hearing them too
I also use iZotope Declick a lot because I hate the "mouth drop" noises a voice often makes, and most compressor can crazy on these things. I kill these out of the signal most of the time. Hate hearing them too
It's not work at all, you just grab a fader and make it sound level to your ear.
You can fine tune the automation if you miss bits when recording fader moves.
If in the end you find riding a fader to be a hell of a lot of work, I don't know what else to say!!!
Selig Audio, LLC
Cool! Gonna check those out.Marco Raaphorst wrote: ↑06 Apr 2018McDSP Moo and Klanghelm MJUC (VST/AU) can do super slow compression which won't sound like a compressor but is more like riding the vocal with a fader. This sometimes works like magic for example on very complex voices with a lot of dynamics.
I also use iZotope Declick a lot because I hate the "mouth drop" noises a voice often makes, and most compressor can crazy on these things. I kill these out of the signal most of the time. Hate hearing them too
There's also a trick where you take a vocal sample, reverse it, and process that way. It's said do be very friendly with transients.
Btw, which exact mouth drop noises do you mean? Something I want to avoid is, when it comes to whispering vocals, there are some nomming-like spitty tongue/mouth noises. xD (Yeah, I know, gotta avoid sugary foods and drinks before performing, that makes a rather thick saliva. Bleh.)
Well sooooryyy! xD I thought you meant, stopping at every tiny bit and to fine adjustments, which could take days.
The way you described it now, sounds much faster, though. But I'm afraid, I would also hurt good dynamics, even if I can adjust. Still better than generic compressors, eh?
- Marco Raaphorst
- Posts: 2504
- Joined: 22 Jan 2015
- Location: The Hague, The Netherlands
- Contact:
General mouth noises.RobC wrote: ↑07 Apr 2018Cool! Gonna check those out.Marco Raaphorst wrote: ↑06 Apr 2018McDSP Moo and Klanghelm MJUC (VST/AU) can do super slow compression which won't sound like a compressor but is more like riding the vocal with a fader. This sometimes works like magic for example on very complex voices with a lot of dynamics.
I also use iZotope Declick a lot because I hate the "mouth drop" noises a voice often makes, and most compressor can crazy on these things. I kill these out of the signal most of the time. Hate hearing them too
There's also a trick where you take a vocal sample, reverse it, and process that way. It's said do be very friendly with transients.
Btw, which exact mouth drop noises do you mean? Something I want to avoid is, when it comes to whispering vocals, there are some nomming-like spitty tongue/mouth noises. xD (Yeah, I know, gotta avoid sugary foods and drinks before performing, that makes a rather thick saliva. Bleh.)
-
- Information
-
Who is online
Users browsing this forum: No registered users and 3 guests