How Spotify Learned to Offer Perfect Music
How Spotify Learned to Offer Perfect Music
Anonim
How Spotify Learned to Offer Perfect Music
How Spotify Learned to Offer Perfect Music

Spotify announced a new Discover Weekly feature a few days ago. Every Monday, the service sends you a playlist with what it thinks is the perfect music for you. We figured out how this function works.

I can be called a music lover of easy virtue. I seem to have tried all the music services that exist in nature. Deezer, Yandex. Music, Google Play Music, Spotify, Apple Music are the ones that come to mind right away. I think there were twice as many of them. After Apple Music came out, I settled on it, but I’m lying if I don’t say that I am still drawn to Spotify.

Despite the bad (subjective) Spotify interface, the high cost of subscription and the hassle of renewing it, I have not finally decided whether to stay on Apple Music after the end of the free period.

And it's all about the recommendations.

Musical Nostradamus

I don’t know why Spotify did it and Apple Music doesn’t do it yet, but the latter offers downright bad music. I liked a lot, no less pressed “I don’t like”, and all to no avail - every second track has to be switched.

Spotify Berlin Office
Spotify Berlin Office

This was not the case with Spotify. The first two weeks I was confused by the abundance of opportunities to find new music. Tabs for charts, new releases, Discover, playlists, hits - even in the last days of using the service, I never started using everything. But even half of these functions were enough to find interesting, and most importantly, suitable music for the occasion.

Believe it or not, the Coffeehouse playlist is really perfect for a cafe, Deep Focus helps you focus, and Beach Vibes makes you feel like you're on a sandy beach.

So I was really excited when Spotify announced Discover Weekly. I knew they would not fail.

Looking for a soundtrack to life

Can you answer the question: "What kind of music do you like?" Most often, the answers range from "a little bit of everything" to "I hate it when they thump and yell." But the only true answer is:

you like good music.

It is different for everyone. For some it's alternative rock and indie, for some it's pop and country … damn, some even like chanson.

IMG_4982 2
IMG_4982 2
IMG_4983 2
IMG_4983 2

This is what the Discover Weekly feature is all about. There is no division into genres, but there is music that you will like. The first thirty-track playlist that came to me has a little bit of everything: there is indie rock, a little chill-out, and quite a bit of EDM. I'm listening to the playlist finishing this article, and out of ten songs I listened to, I didn't want to change one.

Doug Ford, head of Spotify's Music Algorithms Department, should be thanked for the idea and implementation of the function. The service algorithms in real time adjust your preferences after each track you listen to. Moreover, it is not necessary to click like - if you listened and did not miss the track, this will already be reflected in the recommendations.

According to Ford, who was interviewed by Wired, there are thirty-two people in the referral department. This does not mean that they manually select music for the service's 75 million users. However, it is they who are responsible for the algorithms by which the service does it.

What's next

This is just the beginning, according to Shiva Rajaraman, who moved to Spotify after working as a product manager at Google and YouTube. In the future, the service will introduce the Moments feature, and it really blows your mind.

Moments will use the advanced algorithms of the service and select music not only according to your preferences, but also based on what situation you are in:

It's two o'clock in the morning, Friday, and you still listen to music on Spotify? You must be drunk. Here you have Avicii.

Shiva Rajaraman

Spotify will analyze the time of day, weather, your location, and a number of other factors and make recommendations based on them.

Spotify is head and shoulders above Apple Music, Deezer, Google Play Music, and more for recommending and personalizing Spotify. Perhaps competitors will catch up and offer something similar. And it seems that the first one to do it has every chance of winning the streaming service race.

Recommended: