How to Create a Remix – “Work” by Rihanna

Creating a remix is about putting your own spin on an already established idea. When listening to a song on Spotify, I often hear my own melody or expectation of the next verse before the actual verse starts playing.

These differences in expectation mean the original producer is either creating very original tracks, or failing to meet your standards for that type of music.

Rihanna – Work (Hous3 Remix) – 65kbps
Picture of an art piece that looks like a collage of different patterns and iconography. Notably a single eye is at the center.
A remix of another medium

Rihanna – Work (Feat. Drake)

“Work” was a song that always intrigued me as a successful, catchy and fun song, with some very interesting vocal techniques. The chorus of the song has Rihanna singing the word “work” many times over, but not too many to where we forget the whole meaning of the word.

But maybe Rihanna did loose the meaning. In the later portions of the song, “work” looses it’s annunciation and sounds more like a “whur”. This semantic satiation dislocates the idea from the music, and the catchy rhythms take over, making the song very easy to dance to.

Screenshot of the remix with the different layers of audio in a timeline.
Chop it up

The Original Beat

The original production on the Rihanna’s “Work” is a laidback summery vibe with an unrelenting groove. During the chorus, slightly larger bass kicks will hit, but nothing too crazy. The tune is mostly carried by the bass guitar notes that pluck throughout the track, accompanied by some sub bass.

Sub bass, Guitar bass, Melody (mostly bass)

The Remix

With the original beat being on the more subdued side of electronic music, I felt a harder hitting house/dubstep beat would be a nice way to contrast the original. The Hous3 remix includes an electric buzzing bass line, hard hitting kicks and a dubstep-style chorus with explosive wobble.

Structure

In a lot of cases, a remix will take on the structure of the new genre the remix is creating. For instance, the structure of a dubstep remix will usually include a long build up sequence that drops into lovely wobble-fest. But in this case, I chose not to change the structure of the original song at all.

When I was first looking at the pieces of the original song, I started to dissect the timings and flow of the track and found some interesting deviations from typical song structure. These deviations may have been due to a compromise of trying to make the track shorter, or a way of making the track feel faster in getting to the good parts.

Breakdown of the different song elements. Almost half (44%) of the song is chorus and pre chorus. While 36% of the song is the verse.

Typically when writing a song, you want to subdivide everything into 4ths. 4 beats is a bar and 4 bars is a phrase. In this song, the choruses are 6 bars long, adding an additional 2 bars to the typical verse length of 8 bars.

Using this same song structure in the remix was a fun limitation to work around, that helped produce interesting results. Although I wanted to have dubstep elements in the remix, I am not a big fan of the long buildup sections. With the limitation of the song structure in mind, this allowed me to leave the buildup moments to just a couple bars before the chorus.

A Remix is like an Onion

Layering the new elements of the song with the existing elements of the original track can be a great challenge when remixing a track. The only elements I wanted to keep where the vocals. Rihanna sings beautifully and I didn’t want to remove any of the beauty by having my new instruments interfere with her vocal lines.

In early iterations of the remix, the dubstep bass wobbles were harsh and would introduce their own rhythm to the already rhythmic vocal lines. I needed to use wobble elements that reinforced the existing groove of the vocals.

Thinking back to how the repeating “work” line was being reduced to a “whur”, I wanted to use a dubstep wobble that also sounded like a “whur”. In my testing, I found a pre-recorded wobble that fit the criteria and layered in pretty well.

The dubstep wobble used in the remix
How the wobble and the vocals come together

Hard Bass

Next up on the remix agenda is the hard bass. This was an element of the song that I knew I wanted from the start. It needed to be a total contrast from the original melodic bass, and something that felt powerful and electric.

Hard bass from the intro
Hard bass mixed with the vocals, from the first chorus

To make the bass sound powerful, I needed a lot of low-end punch. And for that electric feel, I wanted to add some high-end frequencies as well. Most of the high-end comes from distortion of the bass signal. Importantly, if this bass was going to mesh in with Rihanna’s vocals, it needed to leave some mid to higher frequency ranges a little softer. The resulting bass sound is crisp, with room for vocals to come right through.

The other element of the mix that I wanted to come through was the kick drum. With the hard bass competing for the same low-end space as the kick, I used side-chain compression to automatically reduce the bass whenever the kick came in.

I wrote a blog post about Sidechain Compression here.

Wrap it up

With the vocals, hard bass, dubstep wobble and punchy kick drum in the mix, the remix was simple and effective at conveying the idea of a contrasting-style remix. The rest of production was adding the bells of whistles of a dubstep/electronic song, like the drums and effects.

Because the vocal stems were already well processed, the only techniques I used on the vocals were some compression summing, minor EQ and some reverb. These processing steps were to further mesh the vocals in with the rest of the mix.

You can listen to the remix at the top of this page, although the sound quality is extremely compressed. But until I find a place to host high quality remixes, you can find a slightly better quality version by watching/listening to the music video for the remix below (be sure to play at 1080p quality).

Thanks for reading 🙂

What Is Sidechain Compression?

Before we talk about the sidechain, it’s important to talk about compression. Compression is used heavily throughout the Dreaming City album. Taking inspiration from electronic music group Justice, compression is the glue of the album.

Limiting Compressors

Songs throughout the Dreaming City contain many samples, ambient noises and trippy reverberations. These sounds aren’t necessarily cohesive when played together, but when pushed through the same compressors, the sounds begin to mesh. Two pieces of music or sound may mesh when their notes are in the same key, but also when their effected by processing in the same way.

With a Limiting Compressor, sound begins compressing if the volume reaches a certain level, called the “Ceiling”. Everything past that level is heavily compressed, and is said to be “Limited”. The goal is to compress the sound so it’s not to go louder than the ceiling.

Compressor controller showing the compressor Ceiling set to -0.1dB and a Threshold of -9.7dB. True Peak limiting is enabled.
The limiting compressor used in the song Length of a Moment (Feat. Steven Clapham)

The other parameter is the “Threshold”, which enables gain compensation to boost sound that is quieter than the specified level. This shifts all sounds into a loudness zone, where nothing goes quieter than the threshold and nothing louder than the ceiling.

Sidechain Compression in the Dreaming City

Overall, sounds are all filtered through the master limiting compressor, but individual elements are first affected by their own compression. In this album, all bass heavy synths have sidechain compression applied, to let the drums push through the mix.

Sidechain compression takes one sound as an input and one sound as the effector. The input sound decides how much compression will be applied to the affected sound. In my mix, the input is almost always the kick drum. When the kick drum hits, the compression starts, and the affected bass synth is dampened, allowing the kick to sound large and full of bass.

Sidechain compression in Length of a Moment (Feat. Steven Clapham)

In the preview above, the top module is listening to the kick drum, displaying it’s volume over time. Below is the compressor, affecting the bass synth. On the right side of the compressor, you can see the gain knob ducking in sync with the kick hitting. Listening to the bass, you can hear how the sound seems to pulse with the kick, when usually the bass line would sound like a constant tone.

This form of compression applied to the bass synth is constant across it’s audio level, unlike most compression. This means no matter how loud the bass is, it’s signal will be lowered by the kick drum. If the bass was more dynamic, and included quieter elements, along with louder elements, that worked with the kick, then a more typical compression would be used. to keep the dynamics of a softer sound, but dampen the louder sounds, the side chain would effect the threshold knob instead of the gain knob. Additionally, the ratio would need to be set to something like 10:1 so the sound within the threshold would be 10 times quieter.

What is Paulstretch?

Paulstretch is a program for stretching audio by extreme amounts, to make a 3 minute song 1 hour long (or more or less).

Sample of Paulstretch

In the above example, I recorded myself saying “Hello” and you hear it played back 3 times with different stretch factors. The first hello is the original sample, a 1x stretch factor. The second hello is 2x and the third is 50x.

Paul's Extreme Sound Stretch Interface. Looks like it's from the 1990's.
Paul’s Extreme Sound Stretch Interface

Using this stretching technique, musical notes can be held for an extremely long time, without producing unpleasant audio glitches. That said, when stretching music, results range from beautiful to creepy and even painful. For instance, stretching an electronic song with a big bass heavy drop, can yield some interesting results.

Flume Stretched

In the above example, the song “Say It feat. Tove Lo” by Flume, is stretched 60x at the start of the first chorus. The first thing I notice is the vocals, with a lot of high pitched static. The vocals are singing “say it” while a clap sample is layered with the start of a big synth chord, but the stretching completely changes the recognizability of these elements. With this technique, unique and evolving ambient synth sounds become quite easy to create.

Using Paulstretch in The Dreaming City

The last few Hous3 albums include samples generated with paulstretch and The Dreaming City was no exception. The easiest to hear an example is in the song “Play Me”, as the whole song was created around this technique.

Hous3 – “Play Me”

The background “synth” that plays throughout the track is all Paulstretched audio. Layers of reversed instruments and string section enhances the sound to not be too much of a blur. When parts of the song are stretched with vocal samples, the moving dynamics of the final sample feels full of depth and complexity, when in fact it is a simple stretching of a short soundbite.

Get stretching

Thanks for reading. If you want to hear more Paulstretched music, check out the song “Future Night Club” on Spotify. Or see if you can spot some less noticeable samples throughout The Dreaming City album. And if you have any music featuring Paulstretch, please let me know!
-Hous3

Music Videos of The Dream City

A Dodecahedron surrounded by small Icosahedrons. A screenshot from the music video for "Unremembered" by Hous3
A screenshot from the music video for “Unremembered” by Hous3

With the release of The Dreaming City album, 6 music videos were released along side to accompany each track. The ideology of the album is to allow the listeners to immerse themselves in the soundscapes of each song. Music videos would help keep attention and facilitate immersion.

The ideas behind the videos

Although a live action or storyline-based music video is more typical for mainstream artists to release, a lot of planning and production time would be required if 6 videos were to be produced before release of the album. With this deadline in mind, audio visualizations became a more time-effective way of achieving the goal.

Audio Spectrum of the song "The Length of a Moment (Feat. Steven Clapham)" by Hous3
Audio Spectrum of the song “The Length of a Moment (Feat. Steven Clapham)” by Hous3

The above image is an example of an audio spectrum. This graph shows the audio frequencies and volumes of a short moment from the song. The columns show the sound frequencies from >= 50Hz, up to 20kHz. Their height shows how loud that frequency is playing back at. So in the 50-400Hz area, bass notes are playing, pretty loud at this point. The vocal range is 1.2kHz to 5.6kHz. This graph isn’t as exciting or visually pleasing as a “music video” would demand, so some graphical creativity would come into play.

More than an Excel graph

The first two music videos for “Lake of Dreams” and “I Can’t Make the Dreams Go Away” were created in Adobe After Effects. The idea was to translate the audio spectrum into a pretty moving graphic over a nice backdrop. This effect is seen across many electronic music publishers as a mass produced, quick and dirty music videos, that can be created for $49/month at various rendering service sites. Although the comparison was there, I still wanted to put my attempt forward, to make my own version of these visualization music videos, but better.

Background of a beach in California and foreground of an audio spectrogram in a circle. Screenshot of the music video for "I Can't Make the Dreams Go Away" by Hous3
Screenshot of the music video for “I Can’t Make the Dreams Go Away” by Hous3

The result was fun to realize, as this took me down a path of learning how to create the visualization, as well as digging up old vacation footage of a trip down the California coast. The spectrogram has transformed from a moving excel graph to a circular orb of pulsing rhythm over a beautiful sunset. Starting at 12 noon, the spectrum moves around clockwise from high frequencies (20kHz) to low frequencies at midnight (20Hz).

From 2D to 3D

After the first two videos were complete, I felt the visualizations were underwhelming. Additionally, rendering these, seemingly simple videos, would take hours, due to an inefficient use of processor power. At this point I moved to a different style of visualization in 3D modeling, by using a program called Cinema 4D. The style I was going for was again focused on immersion. Everything in the video needed to be moving to the beat of the song so the soundscape could be easily perceived.

3D rendering software layout of a speaker laying flat down with many balls floating above. Behind the scenes screenshot of the set for the music video "The Length of a Moment (feat. Steven Clapham)" by Hous3
Behind the scenes screenshot of the set for the music video “The Length of a Moment (feat. Steven Clapham)” by Hous3

The most obvious way to convert the audio waves to movement was by creating a virtual speaker that would realistically show what the sound would look like. Though a speaker driver moving back and forth is technically correct, to really show the movement and impact of the sound, I needed something to react to it. By laying the speaker on it’s back and dropping bouncy balls on it’s surface, big spikes in audio levels, like the drop of a song, would translate to an explosion of bouncy balls flying all over the place.

Although I had created similar effects in music videos a long time ago, much of the technique needed to be relearned to realize my vision. Additionally I found many of the interactions and processes could be programmed into Cinema 4D, to facilitate a relatively fast workflow for creating interesting content. The only downside to this technique for creating music videos was the render times, typically taking 8 hours, but sometimes more like 48 hours.

How to render beautiful 3D video

Although I was previously frustrated with Adobe After Effects slow render time of a few hours, rendering something in Cinema 4D that was completely unique and beautiful, seemed to be a worthy wait. The main cause for the massive render times was ray-tracing.

To create a 3D render that could represent a scene from the real world, lighting needs to be physically accurate or it will look like a an animation. Every light in my remaining 4 music videos is accurate to 5 bounces and the results, at high resolution, is beautiful. After building and rendering “The Length of a Moment (Feat. Steven Clapham)” music video, I created and rendered the remaining videos by riffing off the same techniques.

The biggest hurtle was the music video for “Unremembered”, as the render took more than 2 days due to the relatively high rendering fidelity. Although I didn’t use even the medium tiered settings, rending so many frames of video, with so many objects, and tones of reflective surfaces, and so many light sources, made this a big undertaking for my computer. From then on, I optimized for faster render times, as the deadline for the album was approaching.

Geometric shapes dancing around a focal point in the middle of a purple background. Screenshot of the music video for "Dream Printer" by Hous3
Screenshot of the music video for “Dream Printer” by Hous3

It’s show time

With all videos rendered and published to YouTube, all that’s left to do is watch. I’ve linked the playlist of videos from The Dreaming City below. Thanks for reading!
-Hous3

About Hous3

Hous3 is a Canadian producer who creates music and videos to go with them.