The mix of ‘Solo’ is credited to Ralph, Clean Bandit’s Grace Chatto and Jack Patterson, with an engineering credit for Ralph’s engineer Tom AD Fuller. Ralph was the main man at the controls. The hardware stage of his mixing process, as described above, ended with Fuller exporting the analogue mix to Pro Tools.
In the case of ‘Solo’ the Pro Tools mix session ballooned to 80 tracks, consisting of 13 tracks of drums (red), two tracks of bass (light blue), two tracks of guitar (light purple), eight tracks of synths and instrument samples (green and purple), six tracks of Bollywood strings (dark purple), 14 tracks of strings (dark blue), three strings group tracks, including one string reverb record track, eight tracks of vocal chops (green), 10 tracks of Demi Lovato lead vocals, including two lead vocal group tracks, two backing vocal tracks (pink), three auxiliary effect tracks (purple), the aforementioned four group tracks that go to the desk (drums, bass, instruments, vocals, in light blue) and a mix print track.
Ralph’s emphasis on working ‘out of the box’ means only a few tracks have extensive plug-in processing: mainly, the bass, Bollywood strings and lead vocal tracks. There also are very few sends in the session, and only Demi Lovato’s lead vocal tracks have sends to the three aux effect tracks: reverbs from Audio Ease’s Altiverb and Valhalla DSP’s Valhalla Room, plus delay from SoundToys’ Echoboy.
Perhaps Mark Ralph’s most-used plug-in is the Massenburg Design Works equaliser. “The Massenburg EQ is my favourite in-the-box EQ. It is an imitation of the Massenburg 8200 EQ, which I have two of in my rack. It sounds very musical and natural to me, and it’s very quick to use. The general technique to identify a frequency you want to notch out is to narrow the Q down to a minimum, boost the gain to maximum, and sweep along the frequencies until you find the one that bothers you. It’s how I do it on the desk. But with the MDW you literally hit one of the five buttons, and it automatically does everything for you to allow you to find that frequency.”
The increased track count at the mix came about partly because single parts were split across multiple tracks to receive different processing. “I split the acoustic guitars over two tracks, because the intro and middle eight guitars needed very different treatments. The latter has the Valhalla Room for some extra ambience. The main treatments on some of the Bollywood strings come from the Valhalla VintageVerb, the SoundToys Little AlterBoy and the Cableguys ShaperBox. We could have asked the string players to do those pitch slides, but because we wanted things to sound slightly synthesized we used a combination of pitch-bend automation in NI Kontakt, when we resampled them, and Little Alterboy to do this. Instead of going smoothly between notes, it bends up to the notes, and this plug-in is really good for automated pitch shifts like that.
“ShaperBox is VST-only, so I use the Blue Cat PatchWork to get it to work with AAX in Pro Tools. I used the VolumeShaper in ShaperBox, which allows me to create a side-chaining effect without using a trigger. You can draw in the exact curve you want. KickStarter is made by the same company, and also works great, but ShaperBox is more versatile, as it doesn’t only give you volume shaping, but also MIDI shaping, pan shaping and so on.
“The Intro Strings group has the XLN Audio RC20 Retro Color, which is a little find of Jack’s. That plug-in gives you a very nice, natural-sounding array of analogue-type processing. It has noise for general crackle on the track, and you can use it for several different types of overdrive. The overdrive can be very subtle. I used some wobble and distortion on the strings, and I also used RC20 distortion all the way through on Demi’s vocals. It made all the vocals sound like they are slightly overcooked and slightly up and down in pitch a little bit, as though they were coming off a record that wasn’t being played back very reliably. The RC20 also has a very nice reverb.”
On the subject of vocals, Demi Lovato’s vocal tracks are marked ‘AT’, obviously meaning that they’re tuned with Antares’ Auto-Tune. Ralph: “I like that plug-in because of how natural it sounds in graphic mode. Because Auto-Tune’s default page is Auto, people assume that Melodyne is the one where you can draw everything in, and that Auto-Tune just automatically detects. But Auto-Tune has had graphic mode from the very beginning. For me, if I want to tune in a subtle and transparent way where you can’t tell that it’s tuned, Auto-Tune is the best. The down side with AT is that it can be quite glitchy, and it is also prone to CPU overload, but the end result is to my ear much more natural than when using Melodyne. When I listen to the latter, I always feel that there is a synthesized note added in. Obviously, Auto-Tune can’t do polyphonic audio, so if a guitar or piano note in a chord is out of tune I correct that with Melodyne.
“We also have Avid’s Smack! on Demi’s lead vocal. Normally the Waves CL76 is a go-to in-the-box compressor for me, but Smack still gets me an effect that’s most like an outboard compressor. It is really good at detecting pops, and de-essing in a very light way. When I have the vocal on the desk, it will go through my outboard compressor, and sometimes also desk compression. I will err on the side of caution because I know that it will be irreversible by the time I go in the box. When I am in the box and want to hear the vocals really right up front, I will put on Smack! on a more extreme setting, in this case an attack of 5.9 and a ratio of 6:1. The main lead vocal tracks have sends to the two reverb and delay aux tracks lower down the session, where you can see that I completely cut the reverbs and delay in a couple of places, which is a DJ trick: filtering a whole mix in and out.
“All Demi’s chorus vocals are sent to the ‘Demi Chorus Chops Group’ track, on which I have the Waves Greg Wells VoiceCentric. Jack and I were wondering what to do to add some excitement to Demi’s vocals, so we put the Greg Wells plug-in on, and found that it made it sound quite good. So whenever we felt a little unexcited during this mix session, we just reached for the GW plug-in and it made everything sound nice. We used the Greg Wells MixCentric on one of the Bollywood strings tracks. Demi’s chorus group vocals also were treated with the ShaperBox. The mix is set to 37 percent, and it is just to add a little bounce to the vocals. When done in a subtle way, side-chaining on vocals really makes them feel as if they have a little bit more rhythm.”
“As I mentioned earlier, at the bottom of the session are the drums, bass, instruments, and vocals subgroups, and these come up on channels 1-8 on the desk, where I can make further group level and group left/right balance adjustments. You can also see a small adjustment in the level of the vocal group in Pro Tools, where I raise them slightly towards the end. When the mix is completely finished, and I want to print the mix, I go to the SSL computer, with the old green screen and which is still saving mixes on floppies, and I save a total recall session of the desk with the eight faders used. That also stores the VCA 8 settings, which control the amount of SSL mix bus compression. When I print the final mix, I have every single channel muted, apart from channels 1 to 8 of course. I make sure to have no inserts, and no channel EQ or compression on the subgroup channels. It’s also really important to not only mute all of the 40 unused channels at this stage, but also to send them all to the back bus of the Quad channel output on the SSL, because we discovered that otherwise each unused channel bleeds a tiny bit of noise onto the mix bus, even when muted.
“I print back into the main session. At the bottom of the session is the mix print track, which is marked ‘4.2’. In the old days, recalls would take two hours on an SSL so you didn’t do them too often. But today there are endless requests for changes. By the time you have done all those minute changes, you can sometimes end up with more than 50 mix versions. If I look at a mix track, and it says ‘mix 42’ it starts looking really depressing to me. So I tend to go into decimal places, which is nicer for me to look at! So mix 4.2 is quite reassuring: we did it in four mixes. But in actual fact it was 42 times of going backwards and forwards and doing tiny little things.”
The latter, obviously, is one of many strategies Ralph uses to humanise the digital world and the ways of working that it invites, just like mixing things up with analogue and musically with tons of different genres and sounds. The spice and variety that’s the result of his way of working continues to take the world by storm…
“Yes, there is reverb on the bass!” says Mark Ralph. “Jack [Patterson] and I decided to add a short 0.3s plate from the Valhalla VintageVerb. It’s on an aux track next to the bass track, and there’s a send to that aux track, which is automated. There are some places in the track where it’s just bass and drums, and adding some ambience to the bass in those places can work really well. This technique is used a lot in dance music. If it’s done sparingly, reverb can add a little bit of depth and character to bass, without messing with the bottom end. The other thing you can do, although I did not do it in this track, is roll off the bottom end on the reverb return.”
One reason why Mark Ralph evolved his hybrid approach to mixing is to take advantage of the mix bus compressor on his SSL console. “For me, it is a magic button. Many people use the digital equivalent — for example, there’s one in Logic called Glue. Glue is a good way of describing what the SSL mix bus compressor does. When I am working, I like hearing things from the word go with the SSL mix bus compressor on. I always use the same setting: maximum ratio, medium release and attack — two clicks down from the minimum on each — and the threshold is also set to maximum, so it doesn’t actually do any compression until you push the signal going in. The compressor itself is not a recallable aspect of the total recall, so everything on the desk is sent to VCA channel 8, which I use to push the compressor. It is like my input gain, and I never push it beyond about 3dB of gain reduction on the bus compressor.
“From years of experience, whenever I’ve mixed things on an SSL without having the compressor in as I mix and then putting it in at the end of the mix, there are always unintended consequences. It always destroys transients, and there’s pumping, and so on. Whereas, when I mix with the SSL bus compressor in from the first moment, there are no surprises at the end. Everything is the way I want it. One of the great things that has drawn people to using SSL desks over the years is first of all the headroom that you get, meaning that you can drive it right into the red without hearing any noticeable distortion, and second the SSL bus compressor at the end of the chain, which is quite flattering to your sound and gives you more excitement and impact.
“So when I have completed the mix up to around 70 percent, and I am ready to print stems into Pro Tools, my engineer, Tom Fuller, takes out the bus compression and he then prints the stems. Otherwise we would get mix bus compression on every single stem. When the mix is in the computer, and all tracks are sent to the four auxiliary tracks at the bottom of the session, which come back up on the desk on channels 1 to 8, all set to unity, we put the mix bus compressor back in. What I hear should be identical to what I had when I had the mix spread out over the desk. We tested this, and it always is.”
Received wisdom in the pop world is that the ideal song length is about 3:30, and Clean Bandit’s ‘Solo’ is 3:42 long. However, as Mark Ralph opines, the question of song length is a fraught one for any modern producer. “The standard length of a pop song became three and a half minutes because that suited vinyl. When the 12-inch became popular, dance mixes lasted anywhere up to 10 minutes, because that is what you could fit on a 12-inch. Most people are aware of these restrictions, but not many people know about Spotify’s rules, and their implications for music.
“For example, one Spotify rule is that if your song is not listened to for at least 30 seconds, you don’t get paid. This is known as the skip rate. So you have to capture people for 30 seconds or more, or you don’t earn! That has given rise to the very short intros that we are currently seeing, because musicians are being incentivised to try to capture people’s attention for at least 30 seconds. At the other end, Spotify will also not pay anybody who makes a piece of music that is less than 90 seconds. So if your music is 89 seconds in length, you won’t earn a penny, even if you are streamed 1 billion times.
“When I am making songs now, I follow certain criteria that I know work. I know that the short intro is going to help with the streaming skip rate, so I don’t do long intros when I want to have a big Spotify hit. I also don’t bother with long outros because they will cause fewer replays of the track. And if it is a pop song, the format will be very standardised. There’ll be a beginning intro motif, a verse, a pre-chorus (which used to be called a bridge), a chorus plus post chorus, and then you go through the whole 32-bar cycle again, with each section being eight bars long. After that there’s a diversion, some people call it the middle eight, so you take it somewhere else just to give a bit of respite, and finally there’s another chorus.
“Every time I make a song I try to find a slightly different twist on this formula, at the very least not to get bored by always doing the same thing again. But already now, if you give a record company an eight-bar intro they go: ‘Oh, the intro is too long, and it is too energetic at the beginning. That is not what works on Spotify! We want just a nice little stripped-down beginning with hardly anything in it, perhaps four bars, perhaps even two bars, and then we want the vocals to start, because we don’t want people to skip.
“One way of preventing people skipping is to write a song that keeps you waiting for something. A good example is Gotye’s ‘Somebody That I Used To Know’ which is a genius bit of songwriting and production. In general, by the time I get to the middle eight of a song, I don’t want to hear the same thing again, I want to hear the same plus plus plus, and really end on a crescendo. I often do that in the music that I make, and it’s done in ‘Solo’. The coda at the end has the chorus over chords that haven’t been used under the chorus previously in the song, so it’s something you have not heard before, but it also does not go off on a complete tangent that confuses the listener. What is interesting, is that there has been a trend against doing that kind of thing. If you listen to some Drake records, for example, you are not getting much more in the last minute of the song than in the first minute. But it still works, and it still keeps the attention span, and this approach has become immensely popular.
“The other thing that informed our thinking was that the role of the album has changed. Instead of being the starting point of a campaign, or the central point of a campaign, as in the past, it has become the end of the campaign. You release several singles over a period of time, and once you get to about six singles you go: ‘Oh, better release an album now.’ So the album becomes an afterthought, almost, which is a bit sad, because the album used to be the central focus. So what we have done on this album is make a whole extra album on top of all the singles. This is why the deluxe version has 17 tracks. You will be getting a whole body of work in addition to all the hits you’re already familiar with.”