Calvin Harris has been releasing smash hits since 2007. It’s been reported that his DJ sets fetch him over $400k, making him the highest earning DJ’s on the planet. Harris is an incredibly well-rounded musician. He writes, produces and mixes his records as well as singing and playing many instruments.
In this post, I’ll take a close look at his chart-topping song ‘One Kiss’ to see what we can learn from his approach to making music. We can then use these techniques to help guide our decisions to get great results in the studio.
Structure & Arrangement
‘One Kiss’ has a fairly relentless chord progression and driving energy throughout the track, as is typical with house influenced music. Many of the parts are repetitive from one section to the next. The tension is added using filtering, such as a low pass filter on the strings and piano building up to the final chorus.
The repetitive nature of the composition allows for a fairly complicated and fast-moving structure introducing new musical ideas every 8 bars. Notice how the first verse chords are played by the main synth and the second verse chords is played by the piano. This fairly uncommon approach adds an interesting change of timbre whilst helping verse 2 flow effortlessly from the drop.
One Kiss can be broken down into just 10 main parts. When we’re producing and mixing, often it’s better to use fewer sounds rather than stuffing the mix with a load of layers. Too many layers can make a mix sound congested and ends up confusing the listener.
Tonal Balance & Punch Analysis
I’ve compared the drop of ‘One Kiss’ to the drop of 3 other tracks in the same genre of ‘Commercial House’ to see how the tonal balance and punch compares.
Track 1: Cola - Camelphat (House)
Cola is a more typical house track where ‘One Kiss’ crosses over into Pop-House. The Trinity Display in REFERENCE tells us:
- ‘One Kiss’ has 1.4dB less perceived loudness in the low frequencies and has the same punch in that range as ‘Cola’.
- From 200Hz to 2Khz One Kiss has more perceived loudness. This extra presence could be because ‘Cola’ is destined for club play and ‘One Kiss’ is both for club play and radio/streaming play.
- The high frequencies are slightly more prominent in the mix of ‘One Kiss’.
- From the low mids to the high frequencies, ‘One Kiss’ is slightly less punchy.
Track 2: Lullaby - Sigala ft. Paloma Faith (Pop-House)
Sigala’s Lullaby has had less commercial success than ‘One Kiss’ despite both artists having plenty of number ones in the past.
- The low frequencies in ‘One Kiss’ are more prominent than in Lullaby. Sigala’s tracks are always extremely loud (reading 5.3LUFS during the chorus here) To get a track this loud you often have to reduce the low frequencies.
- We can see in the middle band that ‘One Kiss’ measures -4.1SW. This shows that the stereo width of ‘One Kiss’ is much less wide than ‘Lullaby’. In fact it’s much less wide across the whole frequency spectrum.
- The high end of both tracks is almost identical, with a difference of 0.1dB in perceived loudness. If you want a track with great clarity that isn’t too harsh, try both Lullaby and One Kiss as reference tracks.
Track 3: Solo - Clean Bandit (Pop)
Clean Bandit are more on the pop side of the genre spectrum. This comparison is interesting as even when I really zoom into the mix, adding 6 bands in REFERENCE, the tonal balance is extremely similar. The main difference is that ‘One Kiss’ is punchier in the mids. This is the more prominent kick poking through the mix of ‘One Kiss’.
Separation In The Mix
Harris has gone for a solid and centrally focussed mix for most of the elements in his mix. The intermittent introduction of wide strings, piano and brass open up the stereo spectrum to the listener.
Although Harris is known for his records getting a lot of radio play, they also get a lot of club play. Most clubs play audio in mono through their sound systems, so it’s important for club mixes to translate very well when summed to mono. This could be why Harris went for a centrally focussed mix for the majority fo the song.
The infographic shows some overlapping frequencies, such as the main synth and the vocals. Harris has minimized the conflict between these parts by adding more stereo width to the main synth ducking it out of the way of the mono vocal. You can hear it momentarily gets a few dB (decibels) quieter when the vocal comes in. This helps keep the vocal as the focus point. My preferred way of doing this is using a multi-band compressor and ducking the specific frequencies to reduce masking. An interesting way to do this could be to duck the frequencies just in the mid-channel and leave the stereo channels untouched. Below are instructions on how to set that up.
Low Frequencies Analysis
LEVELS shows that there is a little stereo width below 300Hz. Enough to create some nice separation between the kick and the bass, but not so much that phase cancelation occurs when the track is summed to mono. Keeping your low end in the green like ‘One Kiss’ will help your track sound solid both on radio and in a club.
Verse vs Drop Width
As we saw in the mix separation infographic, Harris has gone for a very central mix. However, during the drop, he introduces the piano and brass positioned wider in the mix. This adds a lift to the drop and differentiates the sections.
With over 40 million views in the first few weeks, it was important to make sure the YouTube release sounded as good as possible. From the screen grab below I can see that YouTube turned down the original file by 0.7dB to match it’s streaming target of roughly -13 LUFS (Loudness units full scale). This leads me to believe the original uploaded file was around -11.5 LUFS int and peaking at around -0.9dBTP.
Compared to a lot of other tracks submitted to YouTube, this isn’t a large reduction at all. For example ‘Lullaby - Sigala’ was reduced by 4.9dB. This leads me to believe that Harris and his team decided to submit ‘YouTube optimized’ audio with the music video.
As a result, the dynamic range is 2.7DR more punchy than the promo release for DJ’s to play in clubs.
The results here are a little disappointing. EXPOSE reveals that Spotify had to turn the track down by roughly 5dB to normalize the track to it’s streaming target of around -14 LUFS int. That 5dB of headroom could have been used to introduce more punch into the record like they did for the YouTube version.
DJ city provides audio to DJs for promotional club play, so it’s a very relevant delivery method to measure for this club track. This version is blisteringly loud, hitting a maximum of -5.1 short-term LUFS. Considering audible distortion can start to creep into a mix at -9 short-term LUFS. This version is also peaking at +1.88dBTP (decibels true peak). This might be passable on a club sound system with an excellent digital to analog converter but sounds quite crackly through my laptop speakers.
What Did We Learn:
- The tonal balance of all 4 tracks used in the comparison were extremely similar. We can use these as reference tracks when we want to be sure our songs have a great tonal balance for commercial release.
- Mixing the chorus wider than the verse adds impact and contrast.
- We can still get a huge sounding mix with just 10 elements.
- Optimizing the audio for Youtube can make our tracks punchier.