Once we have created a frame for our mix via solid mix preparation (which is a post for another time), it’s time to focus on the vocals – in many cases the most important instrument in the mix.
The human voice is the most complex instrument we have to deal with, and you have to understand a number of important concepts when dealing with vocals.
A QUICK EXCURSION INTO VOCAL RECORDING
Let’s have a quick excursion at vocal recording. Mix engineers often have to suffer and repair the results of a bad vocal recording, and it’s worth pointing out common mistakes:
1. TOO MUCH TREBLE AND ESSES
Vocal sounds are a matter of trends. For years, the Sony C800G tube microphone has been very popular due to it’s emphasis and fine resolution on higher frequencies. Without a doubt, one of the best mics ever made, but at the same time, depending on the type of vocal, at many times, there is excessive sibilance on the vocaltracks.
A mic that emphasizes the highs is not the main source of trouble here – it’s excessive sibilance pushed into a tracking compressor that creates the problem.
I don’t really recommend a tracking compressor at all when trying to achieve a vocal sound with crisp highs. Along with more organic sounding records, recent years brought a more organic vocal sound back, with less emphasized treble in the vocals.
2. NO POP SHIELD
If you don’t use a pop shield in front of your vocal mic, chances are you will literally have dropouts in your audio signal.
3. RECORDING IN A VOCAL BOOTH
Most small home-made vocal booths are not really properly absorbing the sound across the entire frequency range. All they do is dampening the highs, leaving you not only with a boomy sound, but also with comb-filtering that results in resonances across the mids. Vocals sound better recorded in a larger room, and it is a lot easier dealing with a few more reflections from a large room than removing resonances from comb-filtering.
4. USING A TRACKING COMPRESSOR
Again, if in doubt, don’t use a tracking compressor – it’s a habit from the analogue tape tracking age. Set your mic-pre and tracking levels carefully, and make sure you have plenty of headroom even during the loudest vocal parts of the song.
Tuning vocals is more like fixing the performance than it is mixing, but since it’s technically possible, it’s part of our repertoire of tools and makes us and our work look better when done at the right places.
I you haven’t done it, look into automating the response- time of your tuning plug-in.
When using tuning plug-ins, avoid having all doubles of the same vocal-line tuned. Especially on a backing vocal, if you have two doubled takes, only tune one of them. If you have more than two, only tune half of them, and between the tuned ones, fine-tune them against each other (for example: left BV at -5 cent, right BV at +5 cent).
BUILDING THE LEAD VOCAL-CHAIN
All of the following, please do at a low listening level. I always use my small portable stereo-speakers for that. Start by leveling the untreated lead vocal so that it sits a little low in the track, just loud enough so you can understand the lyrics and follow the melody.
I say this now, and I’ll say it a few more times in this book: keep going back between these building blocks for fine tuning. Just as an example, you will need to re-adjust the sibilance once you’ve added the final „attitude“- compressor. That goes for each building block. Go back and forth between the building blocks and re-adjust. Also, watch your gain staging – make sure the level that comes in at each building block is similar to the output. This chain has 3 dynamics processors with EQs in between, they are kind of „sharing“ the work that has to be done. None of these building blocks need to do the „heavy lifting“ on their own, each has their own role, and less is more.
Bypassing one or more building blocks is an option. Many times you don’t need all of them.
Starting with the untreated, „naked“ lead vocal-track, I want you to first understand the concept of having two types of volume automation on a vocal: before the processing chain, and after it.
This first one evens out the performance of the singer. You could say it creates the continuity of levels in the vocal that you wished the singer had delivered in the first place. This is something that overlaps and goes back to mix preparation, and there is a fair chance that this is already largely fixed. To confirm this, run the mix, go through the song from start to end and find spots where phrases or syllabels of the lead vocals drop in level, or jump out considerably. Correct these phrases in whole db-values – in other words, don’t go too much into details, don’t draw complex automation curves etc., this is just one small step to even out the lead vocal across the song.
Do you feel the sound of the vocal leans towards too much sibilance? Time to use a De-Esser, which is a spe- cialized compressor that has an EQ in the sidechain re- sulting in reducing passages with a lot of treble between 5-12kHz to be reduced in level. The frequency that you want to reduce can usually be set, as well as the threshold (minimum level at which it starts working) and the amount of reduction in dB.
Use an analyzer to determine at what frequencies the esses are jumping out and reduce them considerably.
Alternatively, reducing the esses pre plug-in chain is a more precise way to do this. In Logic I usually isolate the esses to their own regions which I drag to a separate track but same channel strip (so I can select all esses at the same time). I reduce the gain on these regions between 6 – 10 dB, depending on how much the vocal gets com- pressed in the mix. You want to set them low enough to run below the compressor/s threshold (referring to the compressors to be added later on in the plug-in chain). If you can get a De-Esser to achieve the same – use this technique! You can also completely isolate Esses to an- other channel strip and EQ/compress them differently. Same goes for breathing noises.
With the vocal sitting a little low in the track, we now need to develop a feel for what’s missing. We do that by adding frequencies using EQs, starting from the low mids.
A. Broad boost between 200 – 500 Hz
The Pultec MEQ 5 is usually my first EQ in the vocal chain, but you can simulate these (broad) curves with many stock EQs that come with your DAW. I don’t ever go lower than 200 Hz, and occasionally up to 700Hz. The effect we want to get here is that the vocal gets more weight and warmth in the mix. If the vocal is well tracked, it comes with a lot of that quality in the record- ing and you may not need to do anything here. This is why people use Neve 1073s and various tube-based ￼￼equipment (from Tube Mics to EQs/Compressors) dur- ing tracking.
However, a lot of modern vocal recordings sound rather thin, and a nice boost in the low mids fixes it. If you like the character you’re adding with the boost, you can even do a little bit too much off it. We’ll counterbalance it in the next steps. In case the vocal already sounds overly „muddy“ or „boomy“, add a Linear Phase EQ at the beginning of the plug-in chain, locate and remove the frequencies that cause this effect. Watch the inter- dependence of that – once you’ve removed resonances, you have more leeway again to use that broad Pultec- boost again.
B. Tube compressor for tone
After the boost, insert a compressor for tone – we are not after compression for level automation here, but want to add harmonics on top of the low mid boost and create a sense of glue. The classic Fairchild 660/670 works great here, but don’t limit yourself – many compressors can do the job. Just make sure that it’s not noticeable as compression.
Light tape saturation can also work.
In a dense mix, this is were we create the frequencies that make the vocal cut through the rest of the instruments. We can go to extremes here, but before you start playing with the mid boost, set up the compressor that follows it right away. It’s needed to tame the mid boosts as they can get very harsh.
Often, less or no boosting in the mids is needed in less populated parts of the song, but when the vocals are up against a wall of sound, you will need a musically com- posed texture of „cut through“ frequencies there.
A. Various boosts between 1k and 8k (SSL EQ)
This is more complicated to get right, compared to cre- ating warmth. Start with an SSL-type EQ and boost the high shelf at 8k +10dB, then pull back again to 0dB and find a great setting for it somewhere in the middle. Try switching between BELL and SHELF characteristics (BELL will just boost around the set frequency, while shelf also includes all frequencies above. If 8k is boosting sibilance too much, go a tiny bit lower.
Continue by using the HMF band to boost at 4k. And the LMF to boost 2k. Move these around until you find a good balance – but keep in mind, not boosting anything is always an option.
The goal is to create a cluster of mid-boosts that re- ally becomes one colourful and musical texture of mid- boosts between 1 and 8k.
You can achieve good results with stock EQs of your DAW, but there is a reason why SSL EQs are famous for their musicality in the mids. API, Neve works as well. Not a job for a Pultec.
This compressor’s job is to tame the mid-boosts we just created. We want to see it work hard and fast, but it needs to have a lot of musicality to re- duce the harshness of the mid boosts.
My favorite one is the Gates or Retro Sta-Level, but I also like the LA-3A. The LA-2A and Summit TL-100A can work as well – if in doubt, try a few different ones.
The Sta-Level can reduce ridiculous amounts of gain while you won’t notice any pumping, even when com- pressing 20dB or more. If in doubt, set the compressor to less gain reduction. This is one step in the vocal chain where the quality of the compressor makes a difference in how much you can go to the edge.
In general, I love to give recommendations that are in- expensive, but in reality there are situations where high- end equipment has the edge. This is similar to equipment for video and photography. On high quality equipment, ￼￼even a random grainy picture all of a sudden has an ar- ticistic vintage quality about it.
The goal of using this compressor-stage is to create a balance of power between the warmth and presence we created in steps 3 and 4.
Some sort of 1176 compressor can work at this stage, especially if you need the vocal to cut through a dense and/or agressive track. A blue stripe for more harmonics and distortion, or a black face for smoother tones. There are tons of interpretations of the classic 1176 compres- sor – you can try them all.
6. FINAL TONE AND DYNAMICS CONTROL
A final EQ can round off highs and mids. I like a Pultec EQP-1a here, to boost at 20Hz, and Attenuate at 20kHz. Your vocals will sound more analogue when you roll off the top end slightly. Both the boost at 20Hz, and the at- tenuation at 20kHz should not affect the essence of the tone you have created. The boosts can add little bit of weight, and the cut removes top end energy that only hurts at loud volumes.
B. Brickwall Limiter
When optionally using the 1176 in stage 5, it might add some sparkle, appearing as a mixture of transients and added harmonics, so you can use a limiter at the end of the chain to catch occasional peaks that jump out. ONLY catch ocassional peaks, and set the release time higher than 100ms. A modern digital brickwall limiter works well here.
I’ve written another blog-post entirely dedicated to parallel processing, but it’s worth mentioning here – once you really learned to work the vocal chain above, parallel compression will bring your vocal skills to another level:
On most songs that have a certain dynamic, and specifi- cally ballads that start soft and end in a big showdown involving an entire band, orchestra, choir backing vo- cals, etc., I’ve made the following experience:
• the vocal chain for the (light) first verse, that has only a singer with a piano and (maybe) light drums, is found very quickly. A natural vocal sound utiliz- ing 3B and 6A in our chain often works very well.
• as the singer hits higher (and louder) notes, I need to add more elements of the vocal chain to con- trol the dynamics, and also to tame certain reso- nances that come with loud and high notes
￼• on the chorus, especially towards the end, the vocal has to cut trough a wall of sound, and several compressors are working in the chain
• the solution for the dilemma are two separate vo- cal channels: a natural one, and a processed one!
• anything between the two is using a mix of those two channels – by automate both levels you can create the right mixture of them for each section or even single notes
• use this technique when the singer performs across a wide range of notes, switching into falsetto, shouting, hitting rough rock notes, etc.
• just use a separate track for each of those styles, so you can vary settings and create a balance out of them
• more details on this here
VOX AGAINST THE WORLD
I have worked on songs where the backing track was such a big and dense „wall of sound“, that the vocal would not stick out of the mix, regardless the treatment on the track.
There are two secret weapons for that:
1. The Retro StaLevel Gold Edition compressor – while this might be unuseful info for you, as nobody could manage to make a plug-in version of it yet – if you can get your hands on one of these, try the „Triple Mode“ and a mid to fast time constant. It is my ab￼￼solute favorite compressor for lead vocal in a dense mix. The StaLevel is also useful on a lot of other sources, sounds extremely clean and precise, and I’ve seen it compress 30dB without messing with the integrity of the signal!
2. Multiband Sidechaining: there are a bunch of plug- ins that can do this, the most known one is Waves C6 Sidechain. First check where the lead vocal has it’s energy in the mids, this could be anywhere be- tween 1k and 5k. Find the right frequency by sweep- ing through those frequencies with an EQ. Don’t use that EQ on the vocal. Feed the vocal sidechain into the multiband Compressor/EQ and set it so that it reduces exactly this frequency broadly but in a subtle way.
BACKING VOCALS – All Vocals please sing the same song, at the same time!
Backing Vocals are a lot easier to deal with compared to lead vocals – they usually come in groups of 3 and more, and they have to sit behind/underneath the lead vocals. Whereas we strive to create a lot of dynamics, and pre- serve transients on lead vocals, the backing vocals can be more heavily treated. You can even merge several layers of backing vocals into one file for easier handling.
While there’s several software tools that promise to do this for you, you always have to double check if you want a flawless result:
• esses need to sit exactly in sync with where they sit in the lead vocal – otherwise those messy es- ses will jump out of the mix left and right like the snakes on my favourite movie „Snakes on a Plane“; in addition to that, reduce them in level, they must not interfere with the sibilance of the lead vocal!
• breathing noises in backing vocals? pretty useless! You might enjoy the lead singer’s breathing, but not 20 of them! remove them!
BACKING VOCAL CHAIN
As far as processing, backing vocals are a lot more for- giving. I usually copy and paste the lead vocal chain to the backing vocals as a starting point, then costumize on the following parameters:
(the numbers are referring to the plug-in order of the lead vocal chain, which we use as a starting point)
Equally important for the backing vocals – treat same as lead vocals.
Definitely use a De-Esser – backing vocals should have a lot less sibilance compared the lead vocals.
A. The Pultec-boost
Boost at a different frequency relative to the LV- boost. If the lead vocal has a „warmth“-boost at 300Hz, try 200Hz or 500Hz. If you have several backing vocals or LV doubles, try to boost them at different points, so that the boosts are spread across a wide range.
B. Tube compressor for tone
Same as LV, but a bit more gain reduction.
Here’s where we go counter to the boosts of the lead vocal. Reduce the frequencies you have boosted on the LV.
A. Reduce where the LV is boosted (SSL EQ)
B. Compression Same as LV
5. Attitude = Bypass!
6. Final Tone and Dynamics Control
A. EQ. Adjust this while running all vocals. Attenu- ate the treble even more relativ to the lead vocal. Try to boost at 12kHz or 16kHz to create some „shine“ if applicable to the genre.
B. Brickwall Limiter. You can limit a lot more on the backing vocal.