Scott H. Hawley
@drscotthawley.bsky.social
950 followers 760 following 310 posts
Mostly: ML for music production workflows. Professor of Physics & Senior Data Fellow at Belmont University, Nashville TN Head of Research for Hyperstate Music AI. Teacher of audio engineers, Opinions my own. Explainer blog: https://drscotthawley.github.io
Posts Media Videos Starter Packs
Pinned
drscotthawley.bsky.social
The tutorial "Flow With What You Know" was just awarded "Best Blog Post" for ICLR 2025! cf. iclr-blogposts.github.io/2025/about/
drscotthawley.bsky.social
😮🤯 Stoked to see what Anika Nilles & Rush do together! 🤘
drscotthawley.bsky.social
Here's (part of) an answer for Exercise 5: namely a fully convolutional VAE with residual connections, for which the latent space is a "wee image":
gist.github.com/drscotthawle...
drscotthawley.bsky.social
Another good excuse to go visit my colleagues in philosophy. "How many examples do you know of where reasoning is presented in the form of an inner dialogue?"
drscotthawley.bsky.social
New lesson! "An Arsenal of AutoEncoders (AE, VAE, Conditional VAE, VQ-VAE)" Colab link:
github.com/drscotthawle...
drscotthawley.bsky.social
I know people throughout the ages have thought they were living in the last days, but still... I find it unfathomable that any future AntiChrist could be any more AntiChrist-y than our current instance.
drscotthawley.bsky.social
Meanwhile, domain-specific conferences & journals are doing fine. Was your paper accepted then rejected for space considerations or some other bogus claim, and it involves musical audio? Consider submitting to the JAES Special Issue, Deadline Dec 1 2025: users.spa.aalto.fi/vpv/JAES_V73...
users.spa.aalto.fi
drscotthawley.bsky.social
I wonder if all the job postings requiring "publication in NeurIPS, ICLR, ICML, or similar conferences" might have anything to do with the massive glut of submissions to these venues.
Reposted by Scott H. Hawley
euripsconf.bsky.social
And lastly, if @neuripsconf.bsky.social would choose to reverse the decisions on the papers affected by space constraints, we would be happy and able to accommodate their presentation
drscotthawley.bsky.social
Anyone who wrote a negative review of the new Spinal Tap movie did not get the first one either.
Reposted by Scott H. Hawley
drscotthawley.bsky.social
Today I realized I can ask an LLM to generate Jupyter notebooks (w/ text & code) by having it generate Markdown with cell delimiters and then convert the file format via `jupytext`. (Apparently Cursor does this.)
Prompt: "Write a lesson comparing MLP vs CNN for MNIST classification..."
drscotthawley.bsky.social
Exercise for the inquisitive:
"Let's switch roles. Pretend I'm the LLM and you're the user. Send me prompts and I'll try to fulfill them."
drscotthawley.bsky.social
"That is a horrible graph. I cannot take seriously, or even click on, any report from someone who distorts data so badly. 35% is not 3x the size of 23%."
"But it's from Inside Higher Ed."
"🤦‍♂️ We're all fucked."
drscotthawley.bsky.social
Dumbing-down successful!! :-) Made real progress in comprehension and writing-up. Still need to finish the code but the explanation and simplified-math is here for those interested. Hope to have code going in a week or two. colab.research.google.com/drive/14j5vC...
Google Colab
colab.research.google.com
Reposted by Scott H. Hawley
alex-rogers.bsky.social
A really nice description (with code examples) of using @openacousticdevices.info AudioMoth and Raspberry Pi together for soundscape monitoring. towardsdatascience.com/audio-spectr...
Photograph of AudioMoth sensor and Raspberry Pi deployed for soundscape analysis.
drscotthawley.bsky.social
e.g. Haven't made progress in my flow-based inpainting work in a while, and I suspect it's because the papers I'm reading, I still don't really understand them. Need to dumb-down more!
drscotthawley.bsky.social
I often find I can make no progress research-wise until I really, really dumb something down for myself.
Not sure if it's a curse or a blessing, but it does help when (later) teaching the topics.
drscotthawley.bsky.social
I'm now adding "wacky" hidden-text prompt injections into public-facing documents, syllabi, & more. The goal is not to break or "game" automated reviewing systems, but to help differentiate reviews performed solely by LLM from those that included human oversight.