Why You Keep Listening to the Same Songs: A Story About Deepak Chopra, Discovery and Music Mixing
by Boris But, Co-Founder of Mixonset
Here’s a story about Deepak Chopra, discovery, and music mixing, hopefully with seamless transitions in between.
One of Mixonset’s early users, Andrea, told me about her experience finding new songs. She was a huge music lover who spent hours everyday listening to everything from Arizona Zervas to Fleetwood Mac to Disclosure. She was also an insomniac, and found that Indian meditation music would sooth her — so the calming chanting of Deepak Chopra would lull her into sleep.
She came to me one day and told me her Discover Weekly playlist was completely ruined since she started sleeping better. Why? Since she spent hours and hours finding inner peace to Deepak Chopra’s “Spiritual Warrior”, Spotify thought she’d prefer it if half of her songs were all meditation songs!
After laughing for what felt like an eternity, the true insight of the story dawned on me. Music discovery wasn’t just about playing the right song — it was about playing the right song at the right time! And this is simply common sense. Try playing your favorite Death Grips track at a kid’s birthday party.
Music discovery is a struggle
Music streaming services definitely know this too, which is why they’ve created so many pre-curated playlists sorted and arranged by mood, activity, and genre, or even your music taste. But here’s an insight that they still don’t get: music is possibly the most personal and subjective art form in the world. Our music tastes changes from moment to moment in wildly capricious and unpredictable ways. To pre-curate a bunch of playlists is akin to lacing up on a rainy day to catch lightning in a bottle. You might catch something, but most likely it’ll just be a cold.
To pre-curate a bunch of playlists is akin to lacing up on a rainy day to catch lightning in a bottle. You might catch something, but most likely it’ll just be a cold.
What I listen to when I’m happy is likely to be completely different from what you listen to when you are happy. What I listen to when I am working is likely to be completely different from what you listen to when you’re working. To wander down the twisting path of pre-made playlists is ultimately, a fruitless task.
So how do we find the songs that we want to listen to?
There are 50 million songs out there on Spotify, enough chords and melodies to last centuries of continuous listening. But we constantly listen to the same playlists with the same songs, and go back to that nostalgic tune that your ex-girlfriend showed you two summers ago. It feels like we’re stuck in the musical version of Groundhog Day, where we yearn to listen to new music that we like, but we relive the same songs over and over again.
Whether we use Apple Music, Spotify, Deezer, TIDAL or SoundCloud, we are at an unprecedented era of music accessibility. But with the entire world’s historic trove of music at our very fingertips, why do we, strangely, feel bottlenecked when we stream music?
But with the entire world’s historic trove of music at our very fingertips, why do we, strangely, feel bottlenecked when we stream music?
Deezer puts this bottleneck in music discovery mainly down to people being busy, but with so many things in the world competing for your attention, that sounds like an incomplete answer.
The problem lies with how we discover music.
Active Discovery
There’s two main kinds of music discovery: active and inactive.
Active discovery comes more when we’re tired of the songs that we typically enjoy, when we’re reaching out for completely new genres, new sounds, new artists.
Active discovery comes when we’re reaching out for completely new genres, new sounds, new artists.
Personally, I’d argue that active discovery is difficult to replace with an algorithm. It’s hard to tell how we would react to music that is completely different to what we typically listen to. The best way to do active discovery is through suggestions of friends and family. So for the purposes of this article, I will look at inactive music discovery, which is perhaps the biggest challenge for the consumer music experience.
Inactive Discovery
Inactive discovery is when we like a song and want to find more songs that capture that vibe. Since this song is already within our comfort zone, we constantly want to listen to more of it. Most music discovery belongs to this category.
Inactive discovery is when we like a song and want to find more songs that capture that vibe.
Unclear how suggestions are made
Most people attempt to discover more of what they already like everyday — but fail miserably. When we look through our Daily Mix playlist (we use Spotify as an example since they’re actually great at finding the right songs, but terrible at playing them at the right time) our eyes lay upon a series of strange artist names and unknown album art — it’s exciting and new, but we also get a sense of profound confusion.
How are these songs linked to what I enjoy? Why do they think I like these songs? The problem lies in the fact that we don’t know how these songs connect to our music taste.
Suggestions need to be moment-specific
Another layer of complexity in the inactive discovery process is that these algorithmic playlists that are there for us to discover new music is based on what our music taste is as a whole! On the surface level, this may seem like a huge plus. But once we look deeper, we realize that just as important having the right songs is that we need to play them at the right time. Most people wouldn’t play a workout banger the first thing when we wake up. It doesn’t make sense to just chock in all of our possible liked songs in a jumbled up playlist.
Music discovery is just like going to a restaurant. I personally love chorizo, dark chocolate, Japanese ramen, and really aged cheese, but it doesn’t mean I want all those ingredients in the same dish! Even if I like all the separate components in a playlist doesn’t mean that they will sound good together. There still needs to be a high degree of specificity and order — every ingredient can shine when its used at the right time.
The right song at the right time
To carry this analogy, we would want our ingredients to be separated by flavor profiles and flavor combinations, made into different courses so that they all make sense. Making sense — context — is hugely important in the process of music discovery. It’s all about the right song at the right time.
But the process of making groups of disparate, highly differentiated songs mix together is extremely difficult and time-consuming. Think about that Rain Man-like friend you have who is obsessed with his playlist, and spends hours manually arranging the song order of all his playlists to sound good together. Let’s just say it’s not easy.
Music Mixing in Discovery
When my co-founder and I set out to build Mixonset, we weren’t fixated on music discovery as a problem — we just thought we could use Artificial Intelligence to mix a bunch of songs together. But the act of music mixing was more than the sum of its parts: music mixing is all about taking together the songs that we have and figuring out how they fit together. Which song comes first, which song comes second? At which millisecond does the song begin transitioning into the next song without skipping a single beat?
In other words, we are thinking about the time aspect of the listening experience rather than just playing the right song itself. It’s about playing the right song at the right time — no matter what vibe you’re into.
It’s all about playing the right song at the right time — no matter what vibe you’re into.
Music mixing contextualizes your songs
By providing context in a real-time listening experience and understanding that certain songs pair better together, we can increase the likelihood that we would like a new song.
If we extrapolate that same restaurant analogy that I used previously to our approach to music discovery, it will be about creating dishes that make sense. We can throw in ingredients that we enjoy and go well together, like an Atlantic salmon filet, garlic and onions, but also things we’ve never thought about trying before — such as spices like Sumac or even pomegranate. We can add new songs to a mix once we know what the vibe actually is.
This simple paradigm shift has led to some interesting conceptual advancements in how everyone can discover new songs. If we assume that the songs that we already play are songs that we like, we can simply start from there and suggest songs on the go. This is a similar approach that Spotify’s Song Radio takes (it’s such an undervalued feature!), but the problem with Song Radio is that the playlists are still disheveled, disorganized, and random. When the status quo is to ‘shuffle’, music curation becomes absolutely impossible.
Goodbye Shuffle, Hello Smart Mix
What my music-tech startup, Mixonset does, is that we replace ‘shuffle’ with something that we call ‘smart mix’. With ‘smart mix’ we can simply press a button and have contextualized music playing for you — the Holy Grail of having the right song playing at the right time. We arrange the order of the mix based on important musical features like beats per minute, timbre and key. We also automatically generate seamless transitions between two songs with tremendous effect. When two songs cross fade into each other without skipping a beat, subconsciously, we understand that these two songs make sense together.
When two songs cross fade into each other without skipping a beat, subconsciously, we understand that these two songs make sense together.
Making music discovery organic
Mixonset is trying to solve the inactive music discovery problem through the ‘smart mix’ lens. Instead of sifting through a bunch of random playlists to find new music, just play your own playlists and the app mixes in suggested songs in between your own songs. Mixonset essentially expands the collective sound of the songs that you like (your playlist). The ‘in between’ aspect of this is critical, since it makes music discovery an organic process of how we listen.
An early user of ours reported that Spotify helped him find 500 Liked songs over the span of more than 2 years. Using Mixonset, the same user liked 200 Liked songs in just 2 months. That means that Mixonset could be 5 times better at finding new songs than Spotify using this ‘smart mix’ approach to inactive music discovery.
Since we started building Mixonset a year or so ago, we’ve made quantum leaps in our understanding of the music discovery problem. With millions of songs at our fingertips, it only makes sense that the next music revolution is in music discovery — so that we can finally grasp the songs that we truly want to listen to.
This article was written by Boris But, Co-Founder and CMO of Mixonset. Find him on Linkedin or email him at boris@mixonset.com.
To learn more about how Mixonset revolutionizes our music experiences with Smart Mix, install our iOS app for free.
Mixonset is available for Spotify Premium and iTunes MP3 Library. Mixonset Premium lets you find new songs without skipping a beat. Check out how it works here!
