This is a project me and a friend have been working on for the past 8 months, trying to create an easier way to find good songs in the music immensity that Spotify provides.
I am Alin Panaitiu, a fullstack developer (mostly Python, Rust and React), and I quit my job 8 months ago just to try and build something myself. So with the help of Matei Sandu, I started working on Noiseblend, a web app that uses the Spotify API to give you recommendations with the least effort possible.
Of course, I also wanted to make some money off of it, but it seems like there are quite a few impediments and restrictions when you're working with big companies like Spotify.
I admit that while working on this (and a few other side-projects) I think I learned more than I could have learned in 3-4 years of working for someone else.
But still, I wish I could have known that there's no easy way to commercialize an idea that uses Spotify's API.
So Noiseblend is free for all now, I hope it helps you find a few good songs! ^_^
Permission is granted to temporarily download one copy of
the materials (information or software) on Noiseblend's
website for personal, non-commercial transitory viewing
only.
Does that mean I can only open it in one tab? What if I have it open on my desktop - am I then prohibited to open it on my phone?
And how is transient defined? How long can I keep it open in a tab while I browse other websites?
Awesome work. It's so handy for me just now, because my Spotify account got poluted by someone listening to other music on my phone recently.
Does the playing data get back into Spotify? Including the dislikes? Dislikes are not the easiest thing to find/use in the Spotify apps.
Too bad about the startup side. If you have/write a story, I'd like to read it. Spotify and music recommendations are so interesting. It seems like it should be easy enough to recommend edgy things I like, but that doesn't seem to be the case in practice and I wish I knew why that is.
Hehe, I have the exact same problem. My wife is listening to music on my account from time to time and because of that Discover Weekly isn't that awesome anymore.
About the data, everything you are playing goes through Spotify. I have not implemented any playback feature, just a way to control your Spotify devices so that you don't have to go back an forth with the Spotify app. So of course, they have this data.
I can't say the same about the dislikes. They are just a way for Noiseblend to filter your recommendations.
I wish it would work the same on Spotify as lately I've been given some Russian artists to listen to and I don't understand a word from that language. (But it still sounded pretty good)
I think that because of the way Spotify's algorithms work, a dislike button would be quite hard to implement without breaking the ML stuff.
If you like edgy things, make sure to check out the Genres page in Discover. Almost every genre there has a "Not for the masses" popularity option which contains the edge of that genre, things that people haven't gotten to hear yet. Maybe that's for you ¯\_(ツ)_/¯
I'll probably write a piece about all I've learned and been through in the past 8 months, but let me fix the bugs first :)
Is there somewhere that describes what this.. does?
I love music discovery and am very excited to try this, but I'm very unclear if I should be excited.. what does it do different than Spotify, and how does it achieve this?
Ie, I can create a new playlist, add artists, and Spotify will give me new artists it thinks I should add. I can listen to a radio for any song, artist, album, etc. There's quite a few ways to discover music on Spotify, so I'm feeling the need for more information on how this product differs.
Well, everything this does is enabled by the Spotify API. So there's nothing that I'm doing that Spotify can't do already.
The point of this app is to make more easily accessible, some of the lesser known features that Spotify has:
• Tuneable Attributes (such as Danceability and Acousticness) which you can adjust to get songs meeting those values
• Recommendations by multiple artists (which is something like a Radio, but for more than 1 artist)
• Algorithmically created playlists such as
- "The Pulse of [genre]"- containing music that is becoming popular but you probably never heard it before
- "Intro to [genre]" - an attempted algorithmical introduction to that specific genre
• Blends, which are one-time playlists created on the fly based on your listening history and a specific occasion like Workout, Reading, Commute, etc.
For example, did you know that Spotify has all these genres in its database: http://everynoise.com/engenremap.html
Well, after I found out, I was thinking how could I make it easier for people to find their favorite genres and through that, maybe some favorite songs.
So one of the features that Noiseblend has is discovery by genre, which collects your most listened genres and lays them in 3-card view so you can easily find one that you would really like to listen to.
By the way, Noiseblend is completely free, just check it out ^_^
I noticed that the app feels laggier on higher resolutions, probably because of the CSS animations.
I'll probably have to make a compromise between performance and good looking animations.
About the web player, can you please share some ideas about what I can do to improve the workflow?
There is already a way to open a saved playlist in the web player: Click `Open` in the upper left on the playlist page and you should see a `Web` button.
And as far as I know, the web player creates a Spotify Connect device, so if you have a tab open with the web player, you should see the device when you click the big play button on the bottom of the playlist page.
If you're talking about the blends, I admit I didn't think about that. But it shouldn't be hard to implement.
Think lag really turns user off these days so I would advice making some adjustments there and forgo some animations. Might want to have a 'Back' button as well coz it'll be more intuitive instead of clicking your brand's logo. Profile page also doesn't have a homepage button
I was using noiseblend on iOS via chrome browser but it seems to crash too frequently. Might want to check on that.
Oh yes, my oversight. It did open up the webplayer.
14 comments
[ 2.3 ms ] story [ 44.0 ms ] threadI am Alin Panaitiu, a fullstack developer (mostly Python, Rust and React), and I quit my job 8 months ago just to try and build something myself. So with the help of Matei Sandu, I started working on Noiseblend, a web app that uses the Spotify API to give you recommendations with the least effort possible. Of course, I also wanted to make some money off of it, but it seems like there are quite a few impediments and restrictions when you're working with big companies like Spotify.
I admit that while working on this (and a few other side-projects) I think I learned more than I could have learned in 3-4 years of working for someone else. But still, I wish I could have known that there's no easy way to commercialize an idea that uses Spotify's API.
So Noiseblend is free for all now, I hope it helps you find a few good songs! ^_^
And how is transient defined? How long can I keep it open in a tab while I browse other websites?
You can use Noiseblend as much as you want on as many devices as you can handle.
Does the playing data get back into Spotify? Including the dislikes? Dislikes are not the easiest thing to find/use in the Spotify apps.
Too bad about the startup side. If you have/write a story, I'd like to read it. Spotify and music recommendations are so interesting. It seems like it should be easy enough to recommend edgy things I like, but that doesn't seem to be the case in practice and I wish I knew why that is.
About the data, everything you are playing goes through Spotify. I have not implemented any playback feature, just a way to control your Spotify devices so that you don't have to go back an forth with the Spotify app. So of course, they have this data.
I can't say the same about the dislikes. They are just a way for Noiseblend to filter your recommendations. I wish it would work the same on Spotify as lately I've been given some Russian artists to listen to and I don't understand a word from that language. (But it still sounded pretty good) I think that because of the way Spotify's algorithms work, a dislike button would be quite hard to implement without breaking the ML stuff.
If you like edgy things, make sure to check out the Genres page in Discover. Almost every genre there has a "Not for the masses" popularity option which contains the edge of that genre, things that people haven't gotten to hear yet. Maybe that's for you ¯\_(ツ)_/¯
I'll probably write a piece about all I've learned and been through in the past 8 months, but let me fix the bugs first :)
Look forward to any improvements and reading a write-up. Awesome work already, by far. Kudos.
I love music discovery and am very excited to try this, but I'm very unclear if I should be excited.. what does it do different than Spotify, and how does it achieve this?
Ie, I can create a new playlist, add artists, and Spotify will give me new artists it thinks I should add. I can listen to a radio for any song, artist, album, etc. There's quite a few ways to discover music on Spotify, so I'm feeling the need for more information on how this product differs.
• Tuneable Attributes (such as Danceability and Acousticness) which you can adjust to get songs meeting those values • Recommendations by multiple artists (which is something like a Radio, but for more than 1 artist) • Algorithmically created playlists such as - "The Pulse of [genre]"- containing music that is becoming popular but you probably never heard it before - "Intro to [genre]" - an attempted algorithmical introduction to that specific genre • Blends, which are one-time playlists created on the fly based on your listening history and a specific occasion like Workout, Reading, Commute, etc.
For example, did you know that Spotify has all these genres in its database: http://everynoise.com/engenremap.html Well, after I found out, I was thinking how could I make it easier for people to find their favorite genres and through that, maybe some favorite songs. So one of the features that Noiseblend has is discovery by genre, which collects your most listened genres and lays them in 3-card view so you can easily find one that you would really like to listen to.
By the way, Noiseblend is completely free, just check it out ^_^
Yea I remember a while back I was looking into the Spotify API and saw those attributes you talked of. Pretty nifty!
Appreciate your work! Sad that you're having trouble monetizing off of it.
I noticed that the app feels laggier on higher resolutions, probably because of the CSS animations. I'll probably have to make a compromise between performance and good looking animations.
About the web player, can you please share some ideas about what I can do to improve the workflow? There is already a way to open a saved playlist in the web player: Click `Open` in the upper left on the playlist page and you should see a `Web` button.
And as far as I know, the web player creates a Spotify Connect device, so if you have a tab open with the web player, you should see the device when you click the big play button on the bottom of the playlist page.
If you're talking about the blends, I admit I didn't think about that. But it shouldn't be hard to implement.
I was using noiseblend on iOS via chrome browser but it seems to crash too frequently. Might want to check on that.
Oh yes, my oversight. It did open up the webplayer.