Music collections are directed graphs. Let’s treat them that way.

rmn Member Posts: 40 Helper

As a techno DJ who is into the layering and looping, I would like to be able to create associations between tracks and / or loops and optionally define the nature of the relationship, such that I can quickly retrieve relevant items for a given input track during a live session.


For any track in my collection there are several loops and tracks I can layer on top or mix into. When I am exploring my music, I want to store tracks and loops that work together. When I am playing, I want to retrieve items associated with a given input track or loop.

Let's consider the following object to be items in our collection. We explored our collection and found that some items work really well together.

  • B and C go well after A
  • D goes well after B and C
  • E goes well after B and D

Whenever we are playing any of the items in the graph we want to be able to access a list of items considered relevant for that items:

  • Query A returns B and C
  • Query B returns D and E
  • Query C returns D
  • Query D returns E
  • Query E returns None

And what if we are playing A and want to take the audience to E? Ask for the shortest path between A and E.

  • A works with B
  • B works with E
  • A \> C \> D \> E is longer than A \> B \> E

Therefore, if we want arrive at E ASAP, we should play B now.


These use cases are simple examples of what is possible whenever we start thinking of music collections as graphs. Advanced functionality may involve identifying more complex relationships, or outliers perhaps. Although such functionality may very well be considered beyond the scope of an application like Traktor. The main question here is: Are we ready to leave old-fashioned linear playlists behind us and embrace a more capable approach to collection management and exploration?

Looking forward to learning your views on this topic.

6 votes

Closed · Last Updated

This idea did not receive enough support from the community.


  • QWER2580
    QWER2580 Member Posts: 9 Member

    Yeah, this makes a ton of sense to me. The other thing here is that we're all just so used to sites and apps recommending things for us by now. Everything is customized on the fly. With DJing we're still basically in the 00s, adding tracks manually, dropping cue points and labeling tracks. Maybe automating the thinking in this line of work is beyond the intention of NI but someone's going to do it and that will open the door for new people to come in and become a customer.

    It could even be as simple as a little dynamic list at the top of your playlist browser that shows which songs you most often go into from the ones you're playing or recommend something similar but not in your collection via streaming.

  • Patch
    Patch Member Posts: 222 Pro

    We need further expansion of Smart Playlists to be able to implement this properly.

    I really like the ideas in this post, though. Looks like more of a network diagram than a graph…

    But, yeah, better Smart Playlists would allow us to explore this a bit further.

  • Aquarobotics
    Aquarobotics Member Posts: 14 Helper

    I've presented this idea before as 'track linking' where there's a separate browser window for each track that's loaded in a deck. The tracks in that list are tracks that have been linked to the loaded track. This creates a track matrix of reliable mixes and tracks that have association can be quickly found.

This discussion has been closed.
Back To Top