Hey there! I’m Jooyoung Kim, a mixing engineer, music producer, and computer music researcher.
Lately, I’ve been living like a hermit… haha.
A while back, I think I mentioned grumbling about an “Editor Invited” status dragging on for over a month. Well, good news—it’s finally moved to “Under Review”! I’m keeping my fingers crossed because I really need this to wrap up soon so I can secure some thesis funding from my university to keep things afloat. The timeline’s tight, so I’m a bit on edge.
Oh, and I believe I once mentioned working on digitally recreating the SA-2A on this blog.
Here’s how that went down:
- Analog recording grind (plus building a recording program from scratch… ugh). That recording program UI still gives me nightmares. I built it with Qt Creator, and… let’s just say it was a few days of pure torment.
- Training with deep learning using CNN and RNN (LSTM) methods (which meant coding up some PyTorch for the training process…).
- Implementing the compressor with that training data (cue JUCE and Python code-building…).
And, well, it spectacularly flopped. 😭
Just looking at the spectrogram tells you all you need to know, right?
I built a recording program in C++, slogged through endless recording sessions, spent ages training the model, and then tried real-time processing with JUCE—only to realize real-time processing wasn’t happening, so I switched to Python. After pouring over a month into this, it feels like such a letdown.
The RNN (LSTM) approach might still have some potential with more time, but CNN? Total dead end. My original idea would take way too long to execute, so I’ve shelved it for now. That said, the program set itself isn’t half bad, so I’m planning to polish it up and eventually share it on GitHub.
Meanwhile, I’ve pivoted to experimenting with a new topic. For this, I’ve been measuring THD, crosstalk, and frequency response, and I wrote a Python program to store the data and generate graphs.
This one actually turned out pretty well! I’m thinking of sharing some of the code and distributing the program around the time I submit my next paper. It might need a bit more refinement before I consider selling it, though.
Oops, this is starting to sound like a GitHub channel plug, isn’t it?
The first half of this year has been consumed by experiments and coding for my thesis, leaving me with barely any time to work on my own music. 😢 But I’ve got something in the works, and I’m determined to release at least one track next month. I really need to get into a groove and churn stuff out consistently instead of these sporadic bursts… sigh.
I’ve always been drawn to niche things—studying physics as a kid, making music, diving into audio engineering, and now coding for obscure projects. Guess I’m just wired to love the less mainstream stuff, haha.
Anyway, that’s what I’ve been up to lately. With grad school graduation looming, I’m feeling a bit lost and anxious about what’s next. But I’m choosing to believe things will work out and keep pushing forward.
Catch you in the next post! 😊
Jooyoung's Music Blog 