Oh hello there,
You might have forgotten the suggestion a few weeks back for some kind of a workshop. Four weeks is an eternity on the Internet! Well, I haven't. And a few KVR members here haven't either.
So we've started such a workshop.
We've taken the song Flesh And Bone and taken it apart to learn how it was built. Basically I've rendered the individual tracks in two versions - one with full processing and one with no processing at all. Then we're stepping through the process of mixing the song piece by piece. The plan is roughly this:
Week 1: Lead vocals
Week 2: Snare
Week 3: Kick and bass
Week 4: Drum kit
Week 5: Guitars
Week 6: More guitars (!)
Week 7: Background vocals and instruments
Week 8: Panning and reverb
Week 9: Mastering
Each week I'll post the original (raw) version, my processed version and the work of the participants.
We've already started Week 1 (lead vocals), and there are currently two active participants - GeckoEE and Cordelia.
Anyone is free to download the links and have a go for themselves. If you are serious about participating though, you should send me an email to workshop at kimlajoie dot com to get involved. Participants get access to all the audio tracks for mixing and private personal feedback from me on their work.
Ok, enough exposition. Details for Week 1 follow...
-Kim.
KVR Mix Workshop - Week 1: Vocals
-
- KVRAF
- Topic Starter
- 4692 posts since 28 Jan, 2003 from In these very interwebs
-
- KVRAF
- Topic Starter
- 4692 posts since 28 Jan, 2003 from In these very interwebs
Week 1: Lead vocals
Original full mix
Raw track (warning: uncompressed .WAV 26MB)
GeckoEE's version
Cordelia's version
Yonyz' version
Ngarjuna's version
A7's version
Produced track1 and track2
Production notes:
Approach
The two main aspects of the sound we need to focus on are tone and dynamics. Tone is mainly addressed with EQ, and dynamics are mainly addressed with compression. The raw file you'll be working with is named "Raw-Main(incLoud).wav". The reference is split into two files: "Produced-Main.mp3" and "Produced-Loud.mp3". When mixing, I split the track in two because I found I needed slightly different compression settings for the louder singing later in the song. You might find it useful to do the same.
Tone
You'll notice the raw recording has quite a natural tone to it. If we were doing a sparse acoustic song or some quiet jazz, we might not do anything at all to the tone. This song, however, is closer in style to overproduced pop. Broadly speaking, I've done three things to the tone:
1) High pass the low bass out. Fortunately there's not much energy down there to begin with (sometimes low bass can creep in by way of accidental knocks to the mic stand, or the proximity effect of a singer too close to the mic).
2) Reduce the low-mids slightly. This reduces the "fatness" of the vocal sound, and leaves room for the densely-layered background tracks. Generally speaking, the more room you want to make for backgrounds, the more you'll have to dip the low mids. Taken to extreme, you could reduce the vocal to pure articulation in the upper mids and highs. This would make the vocal extremely thin, to the point where it might not even be perceived as a foreground sound anymore- it would be reduced to an articulation detail in the overall mix. At the other extreme, the more low mids you keep, the further back the background instruments have to be (otherwise they'll just pile up into mud).
3) Boost the highs. This exaggerates the articulation in the vocal and gives it a particularly artificial sound. In a dense mix, this helps the vocal stand out (strong higher frequencies are usually perceived as an indicator that a sound source is nearby), and improves intelligibility.
You'll notice that I've left the upper mids alone. This is where the character of the voice is strongest, and I find that modifying this area makes a voice sound unnatural. Boosting it can be useful as a special effect (similar to a "telephone" bandpass filter), and reducing it can be useful to push background vocals further back.
Experiment with different degrees of adjustment, and listen to how modifying the tone changes the focus and style of the sound, and think about how different tones could be useful in different circumstances.
Dynamics
Generally speaking, I apply compression after eq. This way, the compressor is operating on the tone I hear, and isn't responding to bass or low-mid content that I might have removed or reduced. I've written a bit more on the topic in this thread:
http://www.kvraudio.com/forum/viewtopic ... 84#2596984
You'll notice that on this vocal track I've used some pretty heavy compression to even out the dynamics. In the raw recording Maxine has quite a wide dynamic range - not just between phrases, but between words and even syllables. When listening to the track in isolation, it's not so easy to notice it because we're quite used to listening to (and speaking with) a wide dynamic range. When listening in isolation, intelligibility is not a problem because there are no background sounds. However, when we take this dynamic performance and put it in the context of a dense mix such as this one, intelligibility becomes a very big problem. To hear this, simply play the raw vocal track (without compression) in the context of the rest of the mix. Try to find a volume level where you can hear every syllable, but at the same time the vocal is not overpowering or the other instruments area pushed too far in the background. You'll probably find that at a comfortable overall level, some syllables are lost beneath the other instruments, and yet some others poke out and are distracting.
To get an idea of how I'm using compression, load up the raw recording with the produced versions. You should see that the raw recording is kinda "spiky" with a lot of variation in dynamics. By comparison, the produced versions are more much even and consistent. I did this by keeping the attack and release controls of my compressor at medium settings, but increasing the ratio to maximum (15:1 with the plugin I'm using). I then lowered the threshold until the quietest syllables were just about at the threshold, but everything else was above. That way I get the best consistency without raising the noise floor any more than necessary.
Extras - Saturation, Gating, De-essing
I've used a little saturation on the lead vocals. No enough to make them sound distorted, but just enough to give them bit more energy and "hair". Saturation is a way of adding energy (and perceived volume) without increasing the "physical" (metered) level of the sound. It also evens out the dynamics a little by crushing any remaining spikes. The amount and type of saturation you add will depend on the style of music and your personal taste. This particular type of saturation I used gives a very "dry" sound (contrast with the DistSynth or GuitarLines tracks). The exaggerated high frequencies contributes to this effect.
I generally don't use gates unless there's an obvious noise floor problem (or for a special effect). In such a dense mix as this one, and with so much noise, I find that any noise from the vocal track (or any others) generally gets lost in the background. In fact, it adds a little subtle depth to the sound (though it's not necessarily noticeable). Again, this is really a matter for the style of music and personal taste. If I were going for a cleaner, slicker sound for this song I'd probably gate the vocals. As it is, I was going for a dirty, noisy sound, and a small amount of background noise (and bleed!) enhances this effect.
Some pretty heavy de-essing was used on the lead vocals for this song. This was mainly because the exaggerated high end brought out the sibilance more than I would have liked. Regarding signal chain, I generally prefer to place the de-esser after the eq and compression, because both eq and compression can bring out the sibilance and undo any de-essing (particularly if it's a subtle de-esser, as most are).
Week 2 ->
-Kim.
Original full mix
Raw track (warning: uncompressed .WAV 26MB)
GeckoEE's version
Cordelia's version
Yonyz' version
Ngarjuna's version
A7's version
Produced track1 and track2
Production notes:
Approach
The two main aspects of the sound we need to focus on are tone and dynamics. Tone is mainly addressed with EQ, and dynamics are mainly addressed with compression. The raw file you'll be working with is named "Raw-Main(incLoud).wav". The reference is split into two files: "Produced-Main.mp3" and "Produced-Loud.mp3". When mixing, I split the track in two because I found I needed slightly different compression settings for the louder singing later in the song. You might find it useful to do the same.
Tone
You'll notice the raw recording has quite a natural tone to it. If we were doing a sparse acoustic song or some quiet jazz, we might not do anything at all to the tone. This song, however, is closer in style to overproduced pop. Broadly speaking, I've done three things to the tone:
1) High pass the low bass out. Fortunately there's not much energy down there to begin with (sometimes low bass can creep in by way of accidental knocks to the mic stand, or the proximity effect of a singer too close to the mic).
2) Reduce the low-mids slightly. This reduces the "fatness" of the vocal sound, and leaves room for the densely-layered background tracks. Generally speaking, the more room you want to make for backgrounds, the more you'll have to dip the low mids. Taken to extreme, you could reduce the vocal to pure articulation in the upper mids and highs. This would make the vocal extremely thin, to the point where it might not even be perceived as a foreground sound anymore- it would be reduced to an articulation detail in the overall mix. At the other extreme, the more low mids you keep, the further back the background instruments have to be (otherwise they'll just pile up into mud).
3) Boost the highs. This exaggerates the articulation in the vocal and gives it a particularly artificial sound. In a dense mix, this helps the vocal stand out (strong higher frequencies are usually perceived as an indicator that a sound source is nearby), and improves intelligibility.
You'll notice that I've left the upper mids alone. This is where the character of the voice is strongest, and I find that modifying this area makes a voice sound unnatural. Boosting it can be useful as a special effect (similar to a "telephone" bandpass filter), and reducing it can be useful to push background vocals further back.
Experiment with different degrees of adjustment, and listen to how modifying the tone changes the focus and style of the sound, and think about how different tones could be useful in different circumstances.
Dynamics
Generally speaking, I apply compression after eq. This way, the compressor is operating on the tone I hear, and isn't responding to bass or low-mid content that I might have removed or reduced. I've written a bit more on the topic in this thread:
http://www.kvraudio.com/forum/viewtopic ... 84#2596984
You'll notice that on this vocal track I've used some pretty heavy compression to even out the dynamics. In the raw recording Maxine has quite a wide dynamic range - not just between phrases, but between words and even syllables. When listening to the track in isolation, it's not so easy to notice it because we're quite used to listening to (and speaking with) a wide dynamic range. When listening in isolation, intelligibility is not a problem because there are no background sounds. However, when we take this dynamic performance and put it in the context of a dense mix such as this one, intelligibility becomes a very big problem. To hear this, simply play the raw vocal track (without compression) in the context of the rest of the mix. Try to find a volume level where you can hear every syllable, but at the same time the vocal is not overpowering or the other instruments area pushed too far in the background. You'll probably find that at a comfortable overall level, some syllables are lost beneath the other instruments, and yet some others poke out and are distracting.
To get an idea of how I'm using compression, load up the raw recording with the produced versions. You should see that the raw recording is kinda "spiky" with a lot of variation in dynamics. By comparison, the produced versions are more much even and consistent. I did this by keeping the attack and release controls of my compressor at medium settings, but increasing the ratio to maximum (15:1 with the plugin I'm using). I then lowered the threshold until the quietest syllables were just about at the threshold, but everything else was above. That way I get the best consistency without raising the noise floor any more than necessary.
Extras - Saturation, Gating, De-essing
I've used a little saturation on the lead vocals. No enough to make them sound distorted, but just enough to give them bit more energy and "hair". Saturation is a way of adding energy (and perceived volume) without increasing the "physical" (metered) level of the sound. It also evens out the dynamics a little by crushing any remaining spikes. The amount and type of saturation you add will depend on the style of music and your personal taste. This particular type of saturation I used gives a very "dry" sound (contrast with the DistSynth or GuitarLines tracks). The exaggerated high frequencies contributes to this effect.
I generally don't use gates unless there's an obvious noise floor problem (or for a special effect). In such a dense mix as this one, and with so much noise, I find that any noise from the vocal track (or any others) generally gets lost in the background. In fact, it adds a little subtle depth to the sound (though it's not necessarily noticeable). Again, this is really a matter for the style of music and personal taste. If I were going for a cleaner, slicker sound for this song I'd probably gate the vocals. As it is, I was going for a dirty, noisy sound, and a small amount of background noise (and bleed!) enhances this effect.
Some pretty heavy de-essing was used on the lead vocals for this song. This was mainly because the exaggerated high end brought out the sibilance more than I would have liked. Regarding signal chain, I generally prefer to place the de-esser after the eq and compression, because both eq and compression can bring out the sibilance and undo any de-essing (particularly if it's a subtle de-esser, as most are).
Week 2 ->
-Kim.
Last edited by Kim Lajoie on Mon Aug 18, 2008 3:46 pm, edited 2 times in total.
-
- KVRAF
- 6160 posts since 29 Mar, 2003 from Location: Location
bump
....................Don`t blame me for 'The Roots', I just live here. 
-
- KVRAF
- 10234 posts since 17 Sep, 2004 from Austin, TX
Even though I can't bring myself to listen to your most recent pop/trance song (not even by you, sorry!) I simply adore you for sharing your knowledge in such a format.
-
- KVRist
- 371 posts since 27 May, 2007
Does the raw vocal track already have pitch correction applied to it? I hear a considerable amount of artifacting suggesting that.
-
- KVRian
- 943 posts since 15 Mar, 2005
just to add a mix perspective too. i personally wouldnt mix vocals that loud or squshed, they seem to pump a bit wierdly to me too and harsh at times. to give a very(!) rough idea, cutting the full mix about 4db around 2k sounds better to me, far less fatiguing.martian wrote:like the track kim, great vocal!
have you mixed them loud (by my standards) because of taste (sounds good to you)? or to intentionally make the vocals pop out at you on the radio?
another thing, there are some wierd clicks every now and again, an obvious one at 1:08, could be mp3 clipping i guess.
-
- KVRAF
- Topic Starter
- 4692 posts since 28 Jan, 2003 from In these very interwebs
If I recall correctly, that vocal had some rather extreme pitch manipulation applied to it. Not just intonation correction, but significantly changing the melody. Good thing I wasn't going for a sparse natural tone!dysfunctionz wrote:Does the raw vocal track already have pitch correction applied to it? I hear a considerable amount of artifacting suggesting that.
For this style, I actually like the artifacts. Britney Spears here I come...
-Kim.
-
- KVRAF
- Topic Starter
- 4692 posts since 28 Jan, 2003 from In these very interwebs
The whole mix is actually very harsh and scratchy. It's the sound of rampant high passing and saturation. Yes, it was deliberate. Yes, it was a moment in time and I'm a bit over it now[1].martian wrote:just to add a mix perspective too. i personally wouldnt mix vocals that loud or squshed, they seem to pump a bit wierdly to me too and harsh at times. to give a very(!) rough idea, cutting the full mix about 4db around 2k sounds better to me, far less fatiguing.martian wrote:like the track kim, great vocal!
I'm not sure if I've made this clear or not, but I'm not pretending this mix is as an example "reference" or "perfect" mix. I'll be the first to admit it's a flawed mix, and I made choices then that I wouldn't make now. It's quite possible that the people participating in this workshop will actually put out a better mix than the original.
This workshop is not about "this is how you should mix, and you must do it my way". It's about looking inside this particular mix and having a discussion about the production and engineering techniques that create this kind of sound.
Both.martian wrote:have you mixed them loud (by my standards) because of taste (sounds good to you)? or to intentionally make the vocals pop out at you on the radio?
I'm not near my studio at the moment so I can't hear it, but my guess is that it's not an mp3 artifact - it's just a hard edit. There's a few of them around.martian wrote:another thing, there are some wierd clicks every now and again, an obvious one at 1:08, could be mp3 clipping i guess.
-Kim.
[1] The album I'm engineering at the moment is a lot rounder and smoother in sound. It's as much a matter of personal taste as anything else.
-
- KVRist
- 371 posts since 27 May, 2007
Ahh! I didn't want to be blatant about it, but I was thinking, "I'm a rank amateur, and I can make pitch correction sound more natural than that!" It does sound quite mangled, but that's not necessarily a bad thing.Kim (esoundz) wrote:If I recall correctly, that vocal had some rather extreme pitch manipulation applied to it. Not just intonation correction, but significantly changing the melody. Good thing I wasn't going for a sparse natural tone!dysfunctionz wrote:Does the raw vocal track already have pitch correction applied to it? I hear a considerable amount of artifacting suggesting that.
For this style, I actually like the artifacts. Britney Spears here I come...
-Kim.
-
- KVRAF
- 1666 posts since 28 Jun, 2007 from Amazon rain forest
Fabulous reading! Thanks for share your knowledge with us! 
Looking for the next chapters!
Looking for the next chapters!
-
- KVRAF
- Topic Starter
- 4692 posts since 28 Jan, 2003 from In these very interwebs
..
