Just 4 Noise: The Origin Story

Of Drum Machines and Generative AI…

It was a rather dark and grey morning in Berlin, something that wasn’t out of the norm in the middle of February in this area of the world. A day like this in Germany can feel endless, but this particular one sparked something both sad and special. I was going through my usual winter morning routine, which typically included putting on multiple layers of clothing, making a hot pot of tea, and firing up my gear to make some weird noises. I adopted this regimen after two winters in Berlin, learning that I need a warm, creative outlet each morning to give myself the energy necessary to make it through the day. As I switched on my modular rack and admired the blinking lights, I pressed the ‘ON’ button on my reliable drum machine, only to be put off by the lack of a boot-up sequence… That was the day that changed everything.

my broken, ol’ reliable drum machine the day of the incident

As I grew faithful to my hardware, I had completely forgotten all the sadness that comes with sample doom-scrolling. But alas, with no drum machine to accompany my wake-up jam session, I was back at it. Rather than spending an hour jamming and twisting knobs, I spent 45 minutes looking for my sample folders and 15 minutes staring into the abyss, dreading my lack of hard drive organization. Although this jumpstarted what turned out to be another melancholy winter day, that same dreary morning gave birth to a new world of creation.

Outside of the lore and jokes and memes, the idea behind Just 4 Noise came from a real problem in the world of music production. Prior to building anything, we spoke with friends, colleagues, strangers, and even enemies about pain points in their creative workflows, and we noticed a pretty significant trend. As often as people struggle with maintaining a steady stream of inspiration and creativity, people battle with the universe of endless sound worlds available. Simply put: there are too many samples out there. As a bedroom producer playing with digital tools, how can one possibly settle on the perfect kick? I equate it to walking through an American grocery store looking for a simple bag of chips — yeah you can go for the big brand names, but you always feel that there is a hidden flavor gem in that overwhelming aisle, waiting to be discovered and devoured. Too many options and too much data lead to far too much time spent digging rather than creating, or snacking in this example.

I brought these findings to my trusted colleague at the time, Henning, who is a Machine Learning wizard and also generally a great person to talk to about life, problems in the world, and solutions. As I continued to frame the feedback I got as a search problem, he quickly pointed out that the solution to sample digging doesn’t need to be search-oriented at all. And just like that came the conception of Just 4 Noise: a generative AI tool for music producers who hate digging for samples. Henning explained that if we can gather the right set of data, we could create machine learning models that have the ability to generate completely new, high-quality samples that align with the dataset’s core composition. This point of view sparked something in me, and I knew this needed to be built.

Over many cafe coworking sessions, I was not only able to convince Henning to join me on this adventure but also to focus on prompt-based sample generation, a feat that to this day still does not exist in the world of audio sample generative AI. The reason behind this prompt-oriented UI decision came from follow-up conversations I had with those music producers who complained about time wasted digging for samples. They reasoned that they can quite easily describe sounds in their head into words, and if only there were a way to translate those words into a high-quality render, their quality of life would improve tenfold.

Fast forward a few months, and here we are! 9,000 kick drums, a v4 of our baby machine learning model, a few hundred users testing our beta product, and a lot of excitement about the future :) I hope you enjoyed learning a bit about how we got to where we are today. I’ll continue using this forum for humor and all things music and technology.

As always — feel free to get in touch!

- Max (max@just-noise.com)

Previous
Previous

Data, Data, Data