Secrets Of The Mix Engineers: Manny Marroquin

Manny Marroquin's room at Larrabee Studios has an SSL XL 9000 K-series console as its centrepiece.

Eleven mix engineers tried their hand at Kanye West’s ‘Stronger’ before Manny Marroquin finally turned it into a record. He was rewarded with one of the biggest hits of 2007.

Rappers Kanye West and 50 Cent intentionally put their latest albums under starter’s orders on the same date, September 11, 2007, so they could slug it out in the charts, 50 Cent sulkily declaring that he would stop releasing records if he lost the race (a statement he since appears to have retracted). As both artists had enlisted the services of celebrity producer Timbaland, the whole thing felt rather like a family feud gone public. There appears to be no clear winner yet in their who–sells–the–most–albums stakes, but West has had one of the hits of the year in ‘Stronger’, which was a UK and US number 1.

‘Stronger’ is constructed around a sample of ‘Harder, Better, Faster, Stronger’ by French duo Daft Punk, itself based on a sample of a song by ’70s funkateer and keyboardist Edwin Birdsong called ‘Cola Bottle Baby’. Daft Punk strongly quantised the beat and added vocoded vocals, giving the song a Kraftwerk–like robotic feel. They also added a melodic chorus with a descending chord sequence, which was the section West sampled. The American rapper and producer then slowed down and loosened the rhythm, and overdubbed pulsating synths, evocative rapping and singing, transforming a robotic feel into something much more soulful.

Sculpting ‘Stronger’ to its rousing and ear–catching final shape was reportedly not an easy process. Eleven mixers are said to have tried their hands at mixing the song, each of them creating several different versions, but all were found wanting. Enter star mixer Manny Marroquin, who has regularly worked with West for several years, including mixing West’s first solo album, The College Dropout (2004), and West–produced material for Slum Village, Dilated Peoples and Twista.

Over a gestation period of several sessions, beginning at Marroquin’s room at Larrabee Studios in Los Angeles, and ending at Battery Studios in New York, West and Marroquin finally managed to knock the song into a shape they were content with.

Freaking The Sample

“The song was really tricky to mix,” confirms Marroquin. “What helped me was that Kanye and I worked together a lot, so we communicate well and I tend to know what he’s going for. We started with one full 14–hour day here at Larrabee, and after that we did four more sessions at Battery of about three hours each. I was in New York mixing Alicia Keys’ new album, and I could only work on ‘Stronger’ during time off from that. Kanye and I adjusted the arrangements and instrumentation in the mix with each session. A lot of keyboards were added in New York, for instance. Kanye is the kind of guy who likes to search and explore. Maybe he’ll find something that was not there before. So he’s constantly changing things, and you have to be prepared to make changes on cue. Saying ‘Oh, give me two hours to recall that,’ does not exist in his world. You have to be prepared at all times.”

One way in which Marroquin prepared himself was by working with stereo submixes or stems. It’s a preferred working method, because it offers him the flexibility to instantly adjust his mixes, and it also made it easier for him to switch from Larrabee to Battery. “Stems give me a lot of control,” explains Marroquin. “I work on an 80–channel K9000–series SSL at Larrabee, and my approach is still very much a hybrid of analogue and digital. When mixing to stems I print everything with the analogue effects. This way I avoid having to recall things or having to take some of my gear with me. Everything is easily accessible, and I can change things instantly. When we went to Battery, where they have a J–series SSL, I just opened the stem session in Pro Tools, and it included all my analogue effects, and of course also my plug–in setttings. If I needed to change something, I either did a hardware insert or used a plug–in.”

What set Marroquin’s approach apart from the other mixers’ was his decision to put the Daft Punk sample in the foreground. However, this presented him with particular problems. “The way Kanye ‘freaked’ the sample was one of the unique things about that song. He works by sampling a section, and then cutting it up and putting words and phrases and chord changes under different keyboard keys, in this case on the Ensoniq ASR10. He uses samples both as rhythm and melodic parts. The sample drives the song, but because he slightly slowed down the track, and also because of the way he chopped things up and played the notes, there were a lot of glitches. He tends to leave these glitches, because they’re such a pain for the engineer to fix. Since I wanted to feature the sample I had to get rid of the glitches, to make sure that they didn’t detract from the vocal. So I did drastic volume drops of just a few milliseconds on each glitch.

“The other issue about making the Daft Punk sample the focal point of the song was that because it’s so mid–rangey, robotic, and piercing, it tended to dominate everything and take up all the room. The song was about the sample, the vocals, and the kick drums, and I wanted Kanye’s vocal and the lyrics to be the focal point, and for the sample not to get in the way. I couldn’t just pan the sample to one side, because your attention would be drawn to it, so I ended up using a [Waves] Doubler plug–in and a little bit of EQ to spread the sample, ie. to make it sound bigger, and at the same time move it out of the way of Kanye’s vocals in the middle.

“In addition, the song had a lot of elements, over 100, which included a lot of layers, and it was a challenge to make sure that it didn’t sound busy. It was one of these tracks that could easily have sounded way too crowded, with way too much ear candy, and the key to making sure that this didn’t happen was the beat. We really wanted it to work in the clubs, so we worked really hard on the kicks to make sure that the low end was right. The kicks carry the whole bottom-end register; there is a bass, but it didn’t act like a traditional bass, it was more a synth tone following the root notes. Maintaining the warmth and roundness of the song was to me the most important thing. I remember putting a Manley Passive across the stereo bus on the SSL, taking out 2–3dB at 3k, because the more elements we had in the song, the more mid–rangey it was sounding. I prefer Kanye’s work to sound round and musical and warm.”

‘Stronger’

Writers: Kanye West, Thomas Bangalter, Guy–Manuel de Homem–Christo, Edwin Birdsong

Producer: Kanye West

Marroquin: “The first things that happens when I begin a mix is that my assistant prepares the session for me. I’m old school, I treat Pro Tools like a tape machine, so he puts the drums and the whole rhythm section on the left, the vocals on the right, the guitars towards the middle. I have specific channels on the SSL for all instruments, not necessarily because I always want to use the same equipment, but it’s comfortable for things always to be in the same place. After that I listen to the rough mix. This usually gives me a pretty good idea of where the artist and producer want to go. I always think of it as a blueprint. When I begin the mix, I tend to start with the rhythm section, and I’ll be referencing the vocals every 20–30 minutes. With ‘Stronger’ I began the mix by working on the drums, which were very simple: three kicks, one of them an 808, a snare, and a hi–hat. After that I worked on the sample, then the keyboards, and finally the vocals.”

• Drums: Neve 1073, Avalon 2044, Waves C4, Renaissance Bass & Renaissance EQ, SSL EQ, Lexicon 480

“The beat was extremely hard to get. Getting the kicks right was probably the biggest challenge in mixing this song. I think we ended up auditioning a dozen different kick sounds in different combinations, with different EQ, plug–ins, outboard, different filters and triggers, and so on. The funny thing is that in the end we went back to the kick-drum mix we had on the very first day. On the last day, when we were about to print the mix at Battery, and we’re feeling good about it, Kanye says ‘Let me listen to the LA mix one more time.’ I played it, and he said ‘That’s it!’” (There’s a video on YouTube of Kanye, Timbaland and others working on ‘Stronger’ in an unspecified studio, and they’re already visibly struggling with the kicks: www.youtube.com/watch?v=a–ZhIgJbsuY).

Although ‘Stronger’ had relatively few drum tracks, a lot of processing was applied, most notably the multi–band compression and extreme EQ used to craft the four–to–the–floor kick drum.

“In the end we had three kicks. ‘Kick 2’ is the four–to–the–floor chorus kick, and the other two, ‘Kick Soft’ and ‘Kick’, play during the rest of the track. ‘Kick Soft’ is an 808 kick. On ‘Kick 2’ I used the Waves C4, Renaissance Bass and Renaissance EQ plug–ins. The C4 is accentuating highs and lows, in fact it’s adding across the spectrum, the Renaissance Bass is adding at 47Hz, and the Renaissance EQ is rather extreme, adding 16dB at 2k and 17dB at 14k! It’s like ‘Give me more!’, because it’s what’s driving the song, besides the sample.

“On the two non–chorus kicks, I had the same signal paths. Each sound was bussed to a Neve 1073 EQ and then to an Avalon 2044 compressor, and then came up under a separate fader. On the fader itself I added a little bit of SSL compression and SSL EQ, the latter to take some of the woolliness out of the kicks. On the 1073 I added 4dB at 220Hz, and 2dB at 12k. The Avalon was set to a 5:1 ratio, and with the SSL EQ I added a little bit of 50Hz and 300Hz on Kick 1 and took away around 600Hz. The [McDSP] Filterbank E4 is indicated on the soft kick, but it’s bypassed, I didn’t use it.

“The snare wasn’t your typical hard snare; it didn’t have as many low-mids as other snares. I wanted to keep the emphasis on the kicks. When your snare is too big, it doesn’t work with a four–on–the–floor kick. I added 5dB at 100Hz to the snare on the desk, because the snare had a lot of high mids and I didn’t want it too harsh–sounding. Many people will take that frequency out so the snare doesn’t get in the way of the kick, but I wanted roundness on the snare, I wanted to beef it up. You feel it more than you hear it. I also added some 1.3k, and took away some 7k, all using the board.”

• Daft Punk sample: SSL EQ, Waves Doubler, Eventide Eclipse

“The sample was a stereo track, and it goes really low in places, like almost an octave below Kanye’s voice. If you left it alone, nobody would hear it. Because I wanted to feature the sample, I had to ride the levels, sometimes as much as 6–7dB on certain words. You can see the sample rides I did in the Edit window, with most of the peaks in the ‘break’ section. These rides also include the millisecond volume drops to get rid of the glitches. I didn’t use the Filterbank E4 EQ here either. The EQ settings on the sample are the same throughout the song, it was all about the rides. I also added a Rich Chorus effect from an Eventide Eclipse.

Fitting the Daft Punk sample into the track required extensive automation to eliminate glitching and maintain a consistent level. Waves’ Doubler plug–in was used to make it wider and leave space for the lead vocal.

“The other important issue, which I mentioned before, was to get the sample out of the middle, to make space for Kanye’s vocals. I tried two or three different harmonizers and stereo spreaders. I auditioned outboard like the AN2 Spreader, the Vitalizer, a PCM42 delay and an AMS delay, to try to get something that sounded natural, but they all took away from the focal point, which was the vocals. Very often with spreaders, you start to hear a doubling effect, and I didn’t want that. I simply wanted it to sound like the original, just spread out. I ended up using a Waves plug–in called the Doubler, panning things 10 o’clock to 2 o’clock; I didn’t want things panned sharp left and right either, it was just a matter of making the sound less thick in the centre. I printed the Doubler effect with the stem.”

• Keyboards & Guitars: Waves Renaissance EQ, Pultec EQP1A, Neve 33609, Eventide Eclipse, AMS reverb, SSL EQ, Empirical Labs Distressor

“On our first mix day in Los Angeles there weren’t as many keyboards in the mix, a lot were added during the New York sessions, I think using a Juno 106 and a [Clavia] Nord Lead, among other synths. Mike Dean played many of those, and he also played the little bits of guitar that are on the track. I had eight stereo tracks of keyboards in LA, and some of these were replayed in NY. The bass synth [BaSynth01] only plays the root notes, and doesn’t do anything else. The [GARYNE01] track is a Gary Numan sound, it’s a hooky line that only appears in the instrumental section. The ‘4’ is the Renaissance REQ4 plug–in that I used to boost 354Hz and 3364Hz on the Numan sound. Other than that I didn’t do many treatments on the keyboards, they sounded good as they were, so I preferred to leave them alone. It was just a matter of balancing and a little bit of EQ’ing. The only thing I did on the LA keyboards was to sum them all to couple of Pultec EQP1A equalisers, just to add some body, 4dB at 100Hz, and bite, 3.5dB at 8k. On that first day I also added some 3:1 compression with fast release using a Neve 33609. Finally, the main synth line was treated with the Rich Chorus effect from the Eclipse, and I also added some AMS reverb to the keyboards. As for Mike Dean’s guitars, I sent those to a pair of Distressors and added, again, the AMS reverb.”

‘Stronger’ featured numerous keyboard parts, which were not heavily processed, although some plug–in EQ was used.

• Vocals: numerous processors

“I wanted Kanye’s vocal to be really thick and to make sure that you can hear the full frequency spectrum of his voice. I had a lead track for the first and second verse, and a track for the third verse and the vamp, and a track for the ‘I need you right now’ bit, and two tracks with the ‘That don’t kill me’ hook, panned hard left and right, plus an additional track with the tail end of the hook. There were already a lot of heavy effects on the vocals when I received them, mostly filtering.

“The main lead is coming up on an SSL fader, and I’m filtering at 160Hz and 200Hz on the SSL. I’m then sending the vocal to a bus where I’ve inserted a Neve 32264 with a 3:1 ratio and a fast recovery time, and then to another fader, putting a Tube–Tech CL1B compressor on the insert, followed by an Avalon 2055 EQ, followed by a Dbx 902 de–esser. The CL1B was set to a 4:1 ratio with medium attack; the Avalon adds some sizzle way up there at 25k and I’m cutting at 220Hz. There are two compressors because I must have been looking for a speed of attack that the CL1 doesn’t have. The CL1 is a very smooth–sounding compressor, but its attack is not as fast or pronounced. In addition, there are a few bits I delayed with 577ms using the [Line 6] Echo Farm plug–in, you can hear it on words like ‘more’ and ‘haters’.

“For the ‘I need you right now’ section I used an Avalon 2044 on the insert, ratio 20:1 — I really wanted to grab it — fast attack, slow release, and there’s also a [Line 6] Amp Farm plug–in, which is distorting the vocal, emulating as if he’s saying things through a guitar amp: crunchy, distorted, mid–rangey, cutting out all the lows. On the hook section, the ‘That don’t kill me’ bit, I bus the vocals as a stereo unit to two faders, and on the inserts of each channel I have an Avalon 2055 EQ, a Summit DCL200 and a Dbx 902 de–esser. The 2055 has the smoothest high end possible, so I like to use it if I want to take away or add high end, and the DCL is a nice, very smooth, soft–sounding compressor. I used two Dbx 902s to de–ess, in case I added too much high end. I also used a [Waves] Renaissance EQ plug–in, and I used the cue sends, left and right, to go to an AMS delay harmonizer. On the small faders I also had a couple of [Lexicon] PCM42s, set to slap delays, somewhere around 80ms. That creates a little bit of depth, but they’re mixed quite far back, so you don’t actually notice them.

“I also used an SSL side–chain to de–ess the vocals. The 902 is a great de–esser, but it only takes away one frequency. On the board I have a full spectrum, and I use the side–chain to grab other frequencies. I have more control by using both the SSL and the Dbx. What I do on the SSL is an old trick I learned from engineer and mixer Barney Perkins, who used to do all those Babyface records, and it makes the vocal sound. There’s no de–esser that sounds better than this, but it’s a technique that’s complex to explain.

“Basically, I route the vocal signal to two separate channels on the SSL, say channel one and channel two, which are right next to each other. Channel one is my side–chain and works like a de–esser on channel two, which becomes my actual vocal channel. On channel one I’ll set the SSL compressor to a fast attack and also engage a high–pass filter — ie. I filter out all the low end — and do extreme EQ’ing of whatever frequencies I want to take away. I don’t cut, I boost these frequencies; +12dB with a very narrow bandwidth, most often around 6–7k, where most of the ‘esses’ happen. So channel one accentuates what I’m trying to take away! Of course, I take channel one off the stereo bus, so you don’t hear it in the mix.

“I then press the Link button to link channel one to channel two on the right. I engage the compressor on channel two, and what happens is that the more I’m bringing up channel one, the more the compressor ducks the frequencies I don’t want on channel two. So my channel one fader becomes my threshold. I take out the bass frequencies on channel one, because I don’t want them to disappear in channel two. With side–chaining, the frequencies you’re accentuating in your side–chain are ducked in the other channel. I’ve been part of forums during which I’ve discussed this technique, and people found it very hard to understand. Even the people at SSL have a hard time understanding why it works! But it does. The key is that you can use compressors either to increase gain or to create a ceiling — ie. they also work like limiters. And if you were to use straight EQ’ing to take away your ‘esses’, you take all the life and presence out of the vocal. But if you de–ess this way, it retains the personality of the singer; it’s like he or she is in the room with you.”