Like so many of us here at the Kollection, I pride myself in my quest for new music. I actively scour far and wide for fresh finds each and every week. Not because I get tired of my old playlists, but because I get off on the adrenaline rush, goosebumps, and urge to dance when I stumble upon a new song. If I had to imagine what it feels like to actually strike gold, I bet it’s something like finding an absolute belter of a track at 9 in the morning.
Navigating streaming services and the art of finding good music amongst the algorithmic soundscape is becoming increasingly difficult as tech giants make it more and more convenient to eat up what’s served. Do these little audio gnomes calculating my Discover Weekly truly know my eclectic taste in music? How can they possibly weigh my preference for similarity vs diversity? So, what can we do as listeners to go above and beyond in finding that next jam?
Well, to better understand how we can out pizza the Hut, we need an understanding of how these recommendation algorithms actually work. So, in short, the basic operating model behind Discover Weekly algorithms is said to work using three distinct filtering types.
Navigating The Algorithmic Soundscape
This filter cross references behavioral data among similar users to predict your preferences in music. The underlying idea here is to recommend the music similar users like you have been listening to. The problem here is that consumption patterns lead the algorithm to recommend just the popular tracks, failing to acknowledge any actual information about the song itself and leaving the underdog artists in the dust!
Content Based Filtering
utilizes tagged information about the song to make its recommendation. It constantly extracts information from the internet (like from Kollection articles!) to match similar adjectives or nouns describing a song to a recommended song. The obvious issue here is the more a song is talked about online, the more likely it will match tags with one you already like: leaving the underdog artists in the dust, again.
Audio Feature Filtering
looks at waveform layers of individual tracks and compares and contrasts the audio imaging. There’s a lot of techno magic going on here, but basically, algorithms analyze similarities in song’s key, tempo, loudness, or time signature to recommend tracks that match that waveform analysis. It will find songs that are sonically similar to what you already like. Some may find this super cool and helpful, but maybe not if you are looking for diversity in your library!
IS THERE A RIGHT WAY TO DISCOVER MUSIC IN 2021?
I will admit, these techniques are extremely impressive and do in fact yield some amazing finds from time to time. However, there is an underlying problem with the entire system that is overlooked by so many listeners “looking” for new music. That is, an algorithm can only know you so much. Trust your taste in music more than robots! Evidently, everyone will have different expectations on how a recommendation system should and should not operate, but I implore you to take a few of these simple steps to diversify your library!
Utilize Multiple Streaming Platforms
Yes, that means enduring the chaotic mess that is Soundcloud, the deep rabbit holes of Youtube, and even the cringy Tik Tok influencers to find some gems! There is so much music not seeing the light of day through the recommendation algorithms, expanding your search area can only help!
Dive into Artist Playlists
Follow a few of your favorite artists playlists, pay attention to the tracks they share on Instagram, or get real personal by digging through their personal likes on Soundcloud! Not to mention, following highly curated playlists by labels, blogs, and brands (cough cough K WEEKLY!)
Spotify Radio or Soundcloud Stations
Utilize the radio functions of your streaming service of choice. Yes, it does succumb to the flaws of algorithmic recommendation from time to time, but definitely great to find some gems on a long car ride, a run, or with a morning cup of coffee!
Obviously we haven’t been able to do this at live shows for the past year, but don’t be afraid to use shazam while watching DJ sets on Youtube! We all know and love the one guy in the comment section typing out full tracklists.
Share Music with Friends
I can’t stress this enough. No algorithm can know your taste in music more than your friends do. So, text a new song you found this week to a friend! Ask for some music in return, catch up, schedule a coffee date, get married, have kids- just share music people!
At the end of the day, there is no simple or easy solution to having the hottest music library. Truth is it takes a lot of effort, but I promise you it pays off. Listen past the algorithmic soundscape and find some serious gems as so much good music is being released underneath your radar. Here’s to hoping we can get back out there on the dancefloor soon to boogie to all these new tracks you’re finding!