AI Music
I’ve found one of the best way to stress-test AI 'superintelligence' is through music. When I tried tools like Suno, the articulation was shocking.
To an AI, 'taste' is just a statistical possibility. It may not trump true human originality, but it certainly outperforms repetitive thinking. This clears the path for pure curation. I can now spin up 'invisible digital virtuosos' to computationally jam up something new, allowing me to focus on curating genre fusions that are closer to my abstract imagination than anything on the radio
​
It is intelligent enough to change how I create - that is what I find 'super' about it​
The Playlist
Take this with a grain of salt. AI is unconstrained by traditional rules. There is noise amidst the signal. The experience is best enjoyed if you look past the glitches, and focus on the 80% that lands, it sounds amazing.
Lets start with a more traditional one, here is a hindi vocalization imagined on electronica jazz slow rock. The beats aren't regular - but on 7/4 - to create hypnotic movement.
Here's a version with male vocals, same vibe - different articulation
This one's one of my first creations in Suno, to see if it could replicate complex drum patterns and searing electric guitar lead played in a modern jazz way (Just listen to the lead at 1:52 mark)
Now this one's a bit unique, Qawwali's are generally high decibel music, i wanted to know if Ai can help me imagine a mellowed minimal harmonic version of a qawwali'ish vocalization. It kinda did !
African native music has been a favorite of mine, since a long time. Quite underrated tbh. I tried combining two genre's rarely played with african music - orchestral and rock ! So here's some orchestro-jazz ethiopian music with guitar influenced by african picking.
This is a combination you might have never heard - ever ! Hindustani Braj Djent Rock - as weird it may sound, the overlaps created some new textures. (Just skip, if you dont like it ....heh)
Now lets do some experimentation with Punjabi music! (its actually quite great to to experiment with AI as its a very homogenous style of music which AI has had a really large set of data to train itself with). Each of the tracks below is a new kind of cultural clash of music. (let's see which one you like)
Ok, So i am a big fan of carnatic music in general, the rhythmic complexity and calibre it takes to play carnatic music, can shy away jazz musicians at times. However, in this case i wanted to experiment some very complex electronic rhythms (as an anlogy to complex native rhythms) over malyalam vocalizations. I think, the result is literally a new genre or two !
Here's another song where you hear the musical articulations on a modern disco electro-jazz funk version of malayalam vocalization that have very organic deep bass and guitar runs, achieving a life like contrast in music production
Lets hear this one - Tamil vocals with ethnic and ambient jazz rock blend - I was seriously surprised at the part Ai resolved for the high it created around 2:00 minute mark
Finally - a very mainstream hindustani vocalization but re-imagined on complex rhythm (that mainstream music rarely experiments with) thats both organic and electronic - rock and funk - modern and ethnic - old and new - at the same time !
An observation ... from this weird AI music experience !
AI isn't magic - It is simply curative automation !
I view it as a sampling engine for my thoughts, that allows me to sample the realities of an idea without the manual labor of the past. But even automation requires direction - the curative thought is crucial, and hence the human element remains non-negotiable.
​
To orchestrate a scaffolding of a workflow with AI, you must possess the vocabulary of the craft. You cannot guide an agent through a task you can't describe by yourself. Each keyword is a strategic reference point for AI when its automating stuff for you.
To me, the shift is clear : If the present is about coping with technology, the future is about curating with AI
​
​