Doja Cat's 'Say So' was already a worldwide hit when a new version featuring Nicki Minaj rocketed to the top of the charts. One man mixed them both.
'Say So' was both Doja Cat's and Nicki Minaj's first US number one. Doja Cat's original version, released on January 28th, reached number two in the UK and number five in the US. A remix, with Nicki Minaj adding vocals and a rap, was released on May 1st, pushing the song to the much-coveted top spot in the US (different versions of a song that meet certain criteria are counted as one).
'Say So' was written by Amalaratna Zandile Dlamini, Lydia Asrat, Lukasz Gottwald & Yeti Beats. Produced by Tyson Trax.The track also marks the first number one since 2014 for legendary producer Lukasz Gottwald, aka Dr Luke. Known for big hits by Katy Perry, Pink, Avril Lavigne, Katy Perry, Ne-Yo, Shakira, Miley Cyrus and many more, he's spent the last few years embroiled in widely publicised legal problems which appear to have diluted his stream of hits to some degree. The production of 'Say So' is credited to one Tyson Trax, and while neither Luke nor his team will confirm this, it's an open secret that the name is a pseudonym for the great producer.
Man Of Many Hats
Clint Gibbs is Dr Luke's chief engineer and occasional mixer, a role he also fulfils for Prescription Songs, Dr Luke's production company. "I do quality control," explains Gibbs, "final delivery, upload sessions into the archive, and help songwriters and producers with their setups. I also do the IT for the company, email, tech support, and so on. I wear an insane number of hats! But overseeing studio operations and being a mixer is my main one."
Renowned engineer Serban Ghenea remains Dr Luke's go–to mixer, but Gibbs has increasingly taken to mixing the big songs, and mixed both the original version of 'Say So' and the remix, as well as 10 of the 12 songs on Doja Cat's album Hot Pink (2019). To be able to have the time to do all his mixes, and be there for his wife and baby, Gibbs has taken a back seat at Dr Luke's studio and now mostly works at home. Engineering duties at Dr Luke's studio tend to be performed by Kalani Thompson and Tyler Sheppard.
Originally from Las Vegas, where he played in garage bands and was a fan of metal, Gibbs moved to LA in 2007, when he was 21. After studying at the LA Recording School he started work as an intern at Conway Studios in January 2008. It was here that he met Dr Luke, and started working for him and his then-chief engineer Emily Wright, in December 2010.
"Luke had a run–down studio in the Hollywood Hills, but it was super-private and great music was made there. Whenever we worked with big artists and finalised things we went to Conway or the Record Plant. After that Luke had a house in Malibu that we worked out of for a while, and in 2015 we started building our current studio in Hollywood. The studio is pretty simple, with huge PMC speakers. We are big fans of the PMC company, and they built us new speakers for that room. They really wanted to wow us, and to give us the volume we needed. In fact, they built their huge QB1 XPD‑A speakers because of our room. They are as tall as I am! The speakers have plenty of low end, but also are very detailed. I often work on them at low volumes. We also have some nearfield PMC twotwo8 speakers with two sub2s, and some small Genelec 8020s.
Clint Gibbs now mostly works from his home studio, which is based around a MacBook Pro running Pro Tools, a UA Apollo 8p interface, a Grace Design monitor controller, a PreSonus FaderPort for vocal automation, and PMC, Genelec and Beats speakers for monitoring.
"Our mics are the Sony C800G and Chandler REDD Mic, for backing vocals and to record instruments. The C800 is great for pop, because of the detail and the air, which helps vocals cut through the track. But for hip–hop, where the new style is to stand away from the mic and to the side, with chains rattling, it's more of a challenge. Though I like all the extra noises! We want quality stuff going on, so we also have a rack of top outboard, which contains Chandler TG2 mic pres, which Luke has used since I can remember, and I know [producer] Max Martin [with whom Gibbs regularly works] also loves them. They warm up the top end of the C800. We also have the new UA 1176 compressors, which are quieter and more consistent. From there we go straight into Pro Tools. Of course, there are tons of vintage synthesizers and guitars in the studio, and they are recorded via DI and API 3124 mic pres. Once we are in the box, we are in the box.
"I have spent most of the last 10 years trying to keep up with Luke, often going until 4am, but I now do almost all my mixing at my home, also because of the lockdown situation. My studio is pretty bare-bones, with a kitchen table with my MacBook Pro and the latest Pro Tools, my Grace m905 monitor controller, a PreSonus FaderPort single fader for vocal rides, and then I have my [Universal Audio] Apollo 8p, Genelec 8020 and PMC twotwo6 monitors, a Beats Pill for speaker reference, and my Apple Mighty Mouse. I probably do the majority of my mixing here, but I am pretty flexible when it comes to my mixing. I can be anywhere. I will take my laptop, and I will go to the artist's place, or anywhere else. I also have Sennheiser HD650 open-backed headphones. Jerry Finn had them when he worked at Conway, and he's my idol. I really like to check panning on these headphones, and spatial stuff in general. They are easy on my ears, and ideal when my baby is sleeping!"
Watch & Learn
Jerry Finn, who died tragically young at age 39 in 2008, worked with the likes of Green Day, Morrissey and Blink-182. Gibbs also name-checks Emily Wright, Dr Luke, Max Martin and Cirkut as other mentors, and elaborates: "I was pretty green when I started working for Luke. I'd sit on the floor while Emily recorded Britney Spears, making sure Britney could not see me so she'd be more relaxed, and tweaking compressor settings. I learned a lot from watching all these people work, and also from listening to Serban when he did live mixes for us via iTunes stream link. I'm blessed to have worked with so many talented people, and have over the years just slowly worked on my craft. Reading SOS's Inside Tracks also has been a big help! I never asked to be a mixer, but when the call came, it was great."
Gibbs also mixes for several other artists on the Prescription Songs roster and others outside the company, like Azelia Banks, Kim Petras, Allie X, Bonnie Mckee and Emily Warren, to name a few. But for most of his time, Gibbs is on call to mix for Dr Luke. Because he's less and less physically present at the producer's studio, he's worked hard at streamlining the entire process from recording to mixing. He's trained engineers Thompson and Sheppard and set up recording templates for them to record in, and which also make his mixing process easier.
"We have, for example, vocal and comp template tracks that have been tweaked over the years, with EQ, compression,and de-essing plug‑ins, and eighth–note, quarter–note and half–note delays, reverb, and Antares Auto–Tune 5. At Luke's studio we are still on Pro Tools version 10.3.8, HD3, HD I/O — the classic build — because we really like the low latency on Auto–Tune 5, which allows vocalists to hear the effect while singing. However, during this year I plan to upgrade our rigs to the latest version of Pro Tools, with Auto–Tune Pro.
"Templates are important, but the main thing when recording is to have a good front end, with a good microphone, a good mic pre and a good compressor. I am not about adding 10 plug‑ins to a bus to make something sound good. I really believe in making things sound good from the source. If people manage to make it sound great with 10 plug‑ins, awesome. If it works, it works. But I personally believe in less is better, which you will see in my mixes. I don't have plug‑ins all over the session. The productions I get from Luke's camp always sound good, so I really want to just hit the problems and not overthink a mix and re-engineer it."
Clarity & Loudness
With the latter remark, Gibbs hints at an issue that many mixers today are confronted with: with producers getting increasingly good at mixing, artists and producers are more attached to the rough than ever, and this gives mixers less and less space to be creative.
"When I first started, I might have been a little more ambitious and artistic with my mixes," concedes Gibbs. "I think back then I was trying to find my sound. Nowadays my sound is clarity, and loudness. That is what I focus on. But I don't feel restricted, I just want to improve what is there. If I think something different needs to happen, I may hit up the producer and ask if they mind if I try something. But when the artist is very married to the demo, no matter what changes you make, they are not going to like it. So when I mix today, I need to be very delicate. I really want my mix to showcase the production and the song. I don't need to reinvent the wheel. I just want them to go: 'This is everything we wanted, it just has more clarity and is louder than the rough.' I am cool being the guy who improves the rough, I don't feel a need to put my stamp on it, beyond making sure it sounds good. I just take what the producer has done and elevate it."
Clint Gibbs: "For me, the kick, snare and vocals, and the movement that they bring, are the core of the song. If you have them in the right place, your mix is in pretty good standing.
Part of Gibbs' job is to make sure the mixes sound good on all consumer playback media. He mixes mostly on his PMCs, checks on his Genelecs, and, he adds, "I also check my mixes on my iPhone speaker, Airpods, on my laptop speaker, and so on. Very few people are going to listen to the mix in a controlled environment. If it doesn't sound exciting on the phone, I'll revisit the mix and see if I can add something. Most people will first hear the song as a little clip on Instagram announcing its release. If it sounds dope on there, people may go to Spotify or buy the track on iTunes. I do a lot of parallel bass saturation to make sure it translates on phones."
Parallel processing is one of Gibbs' main tools when mixing, and also plays a crucial role in how he achieves maximum loudness. "It starts with initial gain structure," comments Gibbs, "and from there parallel processing to get added loudness to mainly the vocals and drums. At the end there's also a limiter which averages between a 2-3 dB threshold. But I am a really big believer in parallel processing, and it is my main tool for loudness. I match the plug‑ins on parallel tracks to prevent latency between the buses, so I know for sure that I have the exact same amount of delay compensation. I find this gives me the tightest sound that I can get."
Mixing 'Say So'
Gibbs elaborates on how the transfer from tracking to mixing was done. "Usually we record to a two–track of the production, and once we are going to mix, we will hit up the producer to get the stems. In the case of 'Say So', Kalani created a Pro Tools stem session for me, cleaned it up and organised it in the way he knows I prefer, and he then zipped it up and sent it to me. I unzip it and import the session into my system, and the first thing I do is upsample from their 24-bits to 32-bit float. I will bring in my mix template, with all my buses and aux tracks, and then it is off to the races.
Clint Gibbs' mix for 'Say So' included a parallel bus for Waves' Aphex Aural Exciter plug-in.
"I listen to the rough while I import the stems into my template, and then I start listening to the tracks in the session. I listen to the drums, and decide whether they all should go to my drum bus, or maybe there should be a separate hi-hat or cymbal bus, as I have in the 'Say So' session. Then I organise all tracks and apply colour–coding, and create the drums, bass and keys groups, and so on. In this session everything goes to the All bus, which has the Aphex parallel, and then to the Sum bus. Once I have everything organised the way I want it, I do a Save As, which I call 'Ready', and then another Save As, and I start to mix. It means I can always go back to the 'Ready' mix if I want to start over.
"I usually start the mix working on the drums and vocals. I get the drums to where I want them, and then I'll add the vocals. For me, the kick, snare and vocals, and the movement that they bring, are the core of the song. If you have them in the right place, your mix is in pretty good standing. I then add the bass, and after that I'll add the other elements, like guitars, keys and so on, which are filling up the other space. I normally take four hours to do a mix, and then I'll save it, do something else, and check it again later."
Drums
"The only plug‑in directly on an audio drum track is the NI Transient Master on the hi-hat. The two cymbal tracks go to an aux bus, on which I have a three-band UAD Massenburg EQ. The rest of the drum tracks all go to aux buses called Drums and D-Crush, which have identical plug‑ins — as I mentioned, to make absolutely sure there's no problem with plug‑in delay. The first is the Avid Pro Compressor, which is bypassed on the Drums bus. I wanted the drums to go through uncompressed. Then there's an iZotope Ozone 8 EQ pushing at 60Hz. Because there's no 808 in this version, I needed to add some meat to the kick. The Plugin Alliance Black Box adds saturation. I turned the Pentode knob up and had the Mix down at 17 percent. It then goes to the Ozone 8 limiter, which is in analogue mode, and does a lot of shaping of the low end.
The D-Crush bus is for parallel drum compression and thickening, and houses the Avid Pro Compressor, Plugin Alliance's Black Box saturator, and the iZotope Ozone 8 EQ.
"The D-Crush parallel track below has the Pro Compressor activated, doing 12 to 15 dB compression, so it's really crushing! It has the same EQ around 60Hz, but I have taken out some of the top end, because I felt it was getting a little too clicky. The Black Box has the Pentode and Mix settings turned up a little bit more for more dirt, and the Ozone 8 limiter is the same. I blended the Drum Crush bus in with the Drums bus, and I find that you get really huge-sounding drums doing that. This song also sounded very clean, and needed some more grit.
"The outputs of both these drum buses go to the All bus, and they both also have a send to the Aphex bus, though this send is deactivated in the D-Crush bus. Normally all instrument buses go directly to my Sum bus, but in this case I got a last-minute note, asking for the entire track to be hyped up a bit. At this point it was a matter of doing this quickly, and I also wanted to just add a very broad stroke to everything. So I created the All and Aphex buses, and sent tracks to the latter as required. I didn't send the vocals to it, because they were fine, top-end wise. But pretty much everything else, except for the D-Crush bus, was sent to this Aphex bus.
"I tried the UAD Ampex ATR-102 plug‑in on the All and Aphex buses, but turned both of them off, and forgot to remove them from the session. The FabFilter Pro‑Q 3 has a high-pass and notches around 2.2 and 2.7 kHz, just some frequencies that were really piercing. After that the Waves Aphex Aural Exciter is bypassed on the All bus, but really cranked up on the Aphex bus. I automated it, so it's bypassed in the intro and the outro, and adds excitement to the rest of the track. It's the first time I've done something like the Aphex bus."
Instruments
"One bass audio track has the Ozone 8 EQ, adding some 60Hz, and the other has the Klevgrand Knorr Bass Vitalizer. It's the only time I've used the latter. The company makes some pretty cool stuff, like a great cassette emulator and an LP emulator, and this Knorr plug‑in adds a little bit more attack. The tracks go to the Bass bus and its parallel, the Bass Saturation bus, and they have the FabFilter Saturn and the Ozone 8 Vintage Limiter, but I bypassed the Saturn on the Bass bus.
The bass track was treated to a little Ozone EQ boost at 60Hz, and then a dose of Klevgrand's Knorr Bass Enhancer.
"The guitar audio track goes to the Guitars and Guitars Dim buses. This is a rare occasion where I didn't double the plug‑ins. The Guitars track has the API Vision channel strip, adding 240Hz for more body and 5kHz for some bite, and the compressor is off. The Plugin Alliance bx_shredspread adds some more mid-range and widens a bit. The Guitar Dim bus has the UAD Roland Dimension D, adding a chorus effect. This was a creative addition. I felt the track needed a little bit more of a disco feel and that Nile Rodgers chorus guitar sound. Both guitar buses also have a send to the Aphex bus.
Processing the guitars are the API Vision channel strip, boosting at 240Hz and 5kHz, and Brainworx bx_shredspread, adding some stereo width. A UAD Dimension-D chorus effect is applied on the parallel Guitar Dim bus.
"There are no plug‑ins on the keyboards, and the Keys bus has the UAD 1176, but it's bypassed. There's also a Waves F6, and the Waves Spectre, which adds harmonics around 4.3kHz for more excitement. The intro and outro tracks are stems, and whenever I have several instruments printed on the same track, I calle it Music. I have Oeksound Soothe on the outro. It's an amazing plug‑in. It's my go–to for unruly 808s. In this case Soothe got rid of some harshness around 3kHz in the outro. The Music bus has the 1176, but bypassed, and the Pro‑Q 3, and a send to the Aphex bus."
Vocals
"All vocal audio tracks have the same signal path: Metric Halo Channel Strip 3, Avid Bomb Factory BF-76, Avid Dyn-3 De-Esser and in the case of the lead vocals, the FabFilter Pro‑Q 3, all on the inserts, plus sends to the quarter- and eighth–note delays and reverb. These are all part of the recording template, and came from Kalani. I am familiar with the settings, because I worked on them, so I didn't change them. The Metric Halo is doing a high-pass, the BF-76 has all buttons in to give it that pop vocal sound. I added the Q 3, cutting 1dB at 279Hz and at 740Hz. I like to do tons of minimal EQ cuts with many plug‑ins.
All vocal tracks are processed by Metric Halo's Channel Strip 3, Avid's BF76 compressor, and the Avid Dyn-3 de-esser.
"The lead vocal audio tracks also have a send called Voc SC, which goes to the side-chain input of the Waves F6 on the Keys bus. It helps make a little extra pocket for the vocal. It creates dynamic dips in the 900Hz and 2kHz range when she is singing. I also have this F6 on the 808 bus of the remix, doing the same thing. All lead vocal audio tracks go to the LD Vocal aux, which has another instance of the Q 3, taking out quite a bit of low end to make space for the bass, and the McDSP AE600 dynamic EQ, taking out problem areas only when they occur. I do a lot of vocal riding on the lead vocal bus, pushing up specific syllables.
"There's also a track called Echo FX Clean. When I have to mute a word, I will take the word before it and will delay it with some reverb to fill the empty space. The additional plug‑in by Vengeance, the VPS philta XL filter, is doing a high and a low cut to give it more of that radio sound. The other vocal tracks are treated in similar ways as the lead vocals, though the BG2 bus has the Valhalla Vintage Verb, with a 2.4s reverb, Mix at 18 percent. I think that came from Kalani, but I added the FabFilter Pro‑DS, because I wanted to take the 's' frequency out of the reverb. All vocals go to the Vocals bus, which has the UAD 1176, switched off, and the JSTClip for some level boosting. The Vocal Rear track is the parallel, with -20dB compression on the 1176. This is where I get my vocal loudness from.
Global mix processing is provided by Cytomic's The Glue bus compressor, the UAD Chandler Curve Bender EQ, and iZotope's Ozone 8 Limiter.
"The All and Aphex buses go to the Sum bus, on which I have the Cytomic The Glue compressor, the UAD Chandler Curve Bender, and the Ozone 8 Limiter. I actually added The Glue later, because I wanted that pumping sound that Daft Punk gets, where you feed the kick hard into the compressor, and it really makes the compressor pump. I am not a big fan of compression on the stereo bus, I prefer to do my compression in parallel with my sub mixes, but this one needed that little bit of extra squeeze that adds some character to the mix.
"The Curve Bender EQ has a bit of that smiley–face curve, just pushing the lows and the highs. I normally use the FabFilter Pro‑L 2 on the Sum track, but it wasn't doing it for me, because it was rounding off some of the low–end kick transients too much. Instead I went with the Ozone Maximizer, with mode set to Transient, and this enabled me to get it loud without losing transients on the kick. I took the limiter 1dB down for the 24-bit mix I sent to mastering engineer Mike Bozzi, to give him some room to work with."
The Remix
The stems for the beat section over which Nicki Minaj raps.
Download this Zip file of Pro Tools screenshots to view the details of the session.
inside-track-say-so-session.zip
The remix of 'Say So' adds Nicki Minaj on vocals, and there's a section with different instruments, where she raps over 808s and claps. For the rest the instruments and drums are the same as in the original. Gibbs recounts how the remix came into being...
"The Nicki Minaj remix happened very last minute. I got a text from her engineer, Aubry 'Big Juice' Delaine, saying, 'Here's a link for Nicki's vocals.' I opened it up and realised they were for 'Say So'. So I hit up the label and they confirmed that we were doing a remix. As I was mixing, I was getting updated versions of the vocals, verses, and so on. We did the mix, got it mastered, it was all approved, and the day before the remix was supposed to come out, someone wanted to change the beat underneath her vocals. All of a sudden there was an emergency!
"I got an email from Tyson Trax, saying: 'Here are the new beat stems for Nicki.' Among other things they contained some heavy 808s. I had an hour to mix in these new beat stems, because the remix was supposed to be released in Australia a few hours later. I pasted these new beats in just above the vocal VCA of the original session, and in Nicki Minaj's first verse I deactivated the clips from the original beat files for this section. But because the original did not have 808s, they killed my mix bus. I had to automate a new limiter during this part of the remix.
"I put Nicki's vocal at the bottom of the session, and just needed to pair them with the track. I added all plug‑ins on her vocals that you see in the session, but I did not need to do much, because the processed stems Big Juice sent me sounded great! The SoundToys Little Radiator adds a little warmth, and the Metric Halo Channel Strip boosts some top end while also cutting some low end. There are two Waves C6 plug‑ins dynamically suppressing 177Hz and 700Hz, with a -6dB range. There was some build-up in those frequencies that didn't work with the mix. On the inserts of the individual vocal tracks there also is a Waves Q8 doing two sharp 4-6 dB cuts at 10kHz and 11kHz, which was to tame some 's' harshness.
"One of the 808 audio tracks has the FabFilter Pro‑Q 2, with a high pass at 20Hz and a cut around 40Hz, and the Oeksound Soothe on a 'melodic bass control' setting. On the 808s aux bus and 808 parallel bus I have the FabFilter Saturn, Timeless, and the Waves F6. These 808 buses obviously were added for the remix. My approach with them was similar to that of the bass aux. The Saturn is only active on the parallel track, and is a bit dirtier on the 808 than on the bass. The F6 is side-chained to the vocal to suppress the mids of the 808 while Nicki is rapping. The Timeless also is only active on the parallel track, and adds a light chorus that spreads the 808 out slightly."
The 808s originally proved troublesome; Gibbs tamed them with the FabFilter Pro‑Q 3 and an instance of Oeksound's Soothe, set to 'melodic bass control'.
Session Overview
Gibb's original mix session of 'Say So' is a perfect example of his approach, as it features relatively few plug–ins, and a whole host of routing to parallel tracks, which in each case have the exactly same plug–ins on them. The session contains just under 100 tracks, and is well-organised, with extensive colour–coding.
Download this Zip file of Pro Tools screenshots to view the details of the session.
inside-track-say-so-session.zip
At the top is a Meter track, with the Slate Digital FG‑X mastering processor and the Brainworx bx_meter. Underneath that are three mix print tracks, one with the rough mix, one with a mix sent out for feedback, and one 24-bit print that's sent to mastering, 1dB less loud.
Below these tracks the parallel structure of the session becomes immediately apparent. Two parallel aux bus tracks (All and Aphex) with the same plug–ins go to a SUM track above them. There's also a SUMmaster VCA track, and two more VCA tracks (yellow). The VCA tracks are marked by asterisks in their names throughout. Below this master section the instrument group aux tracks begin, with a Drums and a D-Crush parallel, a Cymbal aux, a Bass and a Bass Saturation parallel, a Guitars and a Guitar Dim parallel, and then Keys, Music, EFX Vocal, and two parallel vocal tracks: Vocals and Vocal Rear.
Underneath the aux bus tracks are the audio tracks. Apart from on the vocals, they have hardly any plug–ins. Almost all treatments are done on the buses. Each section of audio tracks has a VCA above. The first section consists of 12 drum audio tracks, with a *drums* VCA above them (all blue). Next are two cymbal tracks (turquoise) and a *cymbal* VCA above, two bass tracks (purple) and a *bass* VCA, two guitar tracks (green) and a *guitar* VCA, seven keys tracks (beige) and a *keys* VCA, two music tracks and a *music* VCA.
The vocal section starts with two audio effect vocal tracks, and an *efxvocal* VCA above, and then a *vocal* VCA and followed by a LD Vocal aux bus, fed by six audio tracks of lead vocals plus one Echo FX Clean vocal track. Next are three backing vocal tracks sent to an Outro Vocals aux, and then 12 harmony vocals sent to a Harmony Vocal aux, and finally a BG2 aux and a Mhm Mm aux, each fed by two audio tracks. At the bottom of the session is an FX Return aux, fed by five aux effect tracks: an eighth-note Waves H‑Delay, a quarter-note and a half-note delay (each with the Line6 EchoFarm), a Reverb aux with the Blue Cat's PatchWork containing a Lexhall plug‑in, and a Backspread track with SoundToys' MicroShift.