The TikTok Tsunami
In the wild and wonderful year of 2020, the phrase “TikTok hit” has become almost synonymous with “viral hit”. A few months ago, we examined the behemoth of a social media app that is TikTok and how it has shifted the music discovery process. We also analysed the insights we received from our AI after analysing 120 TikTok hits.
Today, we will be putting these insights to work as we try to craft the perfect TikTok hit. We will be experimenting to discover a possible AI-driven songwriting process, and how our insights may benefit other musicians.
Our Production Goals
To recap, here are some of the criteria we want to achieve when making a TikTok hit. These insights were derived from the metadata tags of 120 TikTok hit songs, tags generated by our AI.
Key insight: Hip hop makes up a staggering 50% of all TikTok hits, so our song should be a hip hop track.
Key insight: Majority of the hit songs lie within either 60 to 80 bpm, or 105 to 130 bpm, likely because these bpm ranges are suited to dancing. We should ensure our track is within or near these BPM ranges.
Key insight: The moods we should be aiming for are (in order): Powerful, Energetic and Quirky. We should avoid negative moods such as Sadness, Neutral or Scary.
90% Of The Job Is About Finding the Right Samples (Production Process)
Now that we have no qualms about what we are trying to accomplish, let’s begin our road down TikTok fame (or degeneracy). For the purposes of this experiment, we’ll be focusing entirely on the beat, ignoring some of the arguably more important parts of hit songwriting, such as the melody and lyrics of the vocal topline. That could be part of a different blogpost in the future.
Before we jump into the process itself, let’s breakdown the key components of a hip hop / trap beat:
- Kick: In more modern hip hop, usually a distorted 808, could double as a bass
- Bass: Together with the Kick, this defines the underlying groove of the song
- Snare, hi hat and other percussion
- Melodic element: This could be anything from a traditional synth key riff to some obnoxious siren / noise
Now, I can’t pretend to know what other producers do, but my workflow for hip hop / trap is pretty simple: focus on the melodic element first. That element will define the mood, and then the rest of the components will naturally fall into place. Since we’re aiming for something powerful and quirky, that should help us filter down candidate sounds.
In hip hop, the melodic element usually plays unobtrusively in the background, and is usually something higher pitched that drives the song forward without masking the vocals (hence the tradition with synth keys / organs). Since we’re trying to make a TikTok hit, not winning an Academy Award for Best Original Score, quirky doesn’t mean trying to reinvent the wheel, but being distinctive enough to not be drowned out in the sea of sound-alike rap beats. And thus, when looking for samples, we will keep in mind that we’re looking for something that could work in a similar fashion.
After some hours of digging through sample libraries, and with the hip hop attitude of “represent” (drawing from the culture of your region, which in our case was Southeast Asia), I chanced upon some spicy gamelan samples that were a perfect fit.
With our core melodic element out of the way, finding drum samples that could pair well with our gamelan loop is the next step. Once again, with “powerful” in mind, I went with samples that are more distorted to bring out a rougher vibe, and decided on a distorted 808 for the bass-and-drum part, and some harsh sounding hi hats and snares to round out the drumset.
Now that we have all our parts in place, putting everything together in a nice little package will be left to your imagination, with the end result sounding something like this.
My creation of the "perfect" TikTok hit (produced by brittle bear)
So, How Did We Do?
How did our little beat sample do? We ran that snippet through our AI, and this is how we scored:
Electronic is a surprise, but since hip hop is present, there is no problem.
Our energy levels being medium could be a problem, but can be fixed by introducing more varied percussion, and also having an angry rapper to bring it up to high energy..
For Mood, we’ve mostly hit the mark here, but I suppose the powerful could have a higher confidence value than the quirky. But otherwise it’s right where we want it to be.
The emotion is tagged as positive. We’ve avoided giving our track negative moods, which is exactly what we want.
The track is A# major at a tempo of 86 BPM with a small variance. The BPM is defined by the gamelan samples we chose, being slightly out of the optimal range of 60-80 BPM, which can be easily solved by dialing down the bpm in our DAW.
As our experiment has demonstrated, AI has the potential to completely upend traditional songwriting and production workflows, and reshape the music industry. If you’d like to learn how to incorporate AI into your songwriting and production workflow, you can try out our Musiio App. In there, you can tag your favourite songs (or your own songs) and use the audio search to comb through your catalogue for insights. Write to us at email@example.com to find out more.
About the Producer and Author
This track and post were created by Shaun Goh, who produces under the name brittle bear. brittle bear is a Singapore-based electronic music producer and songwriter who seeks to bring a cathartic asian flare to bog-standard electronic music.