Notable logs of mine:
Research oriented zet:
Thoughts related to creative process:
General updates related to my main personal wiki:
A re-#introduction
Hi, I'm Paul. I'm a computer musician and researcher, currently thinking about lyricism what it means for a computer to sing. I explore quirky vocal synthesis DSP algorithms and try to control them in interesting ways.
As a musician-turned-programmer, one of the things I try to do is build resources for topics related to audio programming and DSP that I wish I had when I was starting out. My aim is to empower digital musicians to "dig in" more and craft their own toys and tools for their work.
I also bake a lot. Expect many pictures of bread and croissants here, among other things.
A brief guide to me:
I "organize" many projects and ideas on my personal wiki:
I talk about what I'm currently working on in my "now" page:
I keep a list of projects I've done:
https://pbat.ch/wiki/projects/
I use "metapages" to better organize thoughts and wiki pages:
https://pbat.ch/wiki/metapages/
For more research-y stuff, I maintain a zettelkasten-wiki that functions as a knowledge base:
https://pbat.ch/brain/
https://pbat.ch/brain/now/
https://pbat.ch/brain/wiki_index/
I cannot recommend the AudioQuest Dragonfly black as a suitable USB DAC for music production.
I figured out today that it's been shaving off transients, to the point where drum synthesis sound design is practically impossible. I guess it might be the preamp?
Shame, cuz I really liked the form factor of it.
While I've put a fair amount of time studying vocal synthesis techniques at a DSP level, there comes a point where you need to actually control the thing. Those systems are what interest me more, and where I expect most of my time and energy to be spent. Many of the techniques I'm building have applications for wavetable morphing as well.
Musically speaking, it doesn't matter how complex or cool the DSP algorithm is. How it gets performed (by computer or human) is far more important.
At a very high level, (morphing) wavetable synthesis and vocal synthesis have a lot in common. Both produce complex timbres through a mechanism that navigates/interpolates through a state space over a period of time.
In vocal synthesis, the states are phonemes. In wavetables synthesis, the states are wavetables.
I guess you could go one step further and say both are forms of spectral morphing synthesizers
This new publication is looking for articles about computer history! I think there may be some folks here with some stuff to say about such thing -- https://history.computer
Fountain Pen Gushing
This Kaweco Lilliput fountain pen I got a few days ago is perfect for my needs. It's compact AND it writes a nice bold line. It comes in stainless steel which is a nice plus.
Seriously, there doesn't seem to be any other pen that comes close to compactness and writability, fountain or otherwise. I would love to be corrected here.
The Fisher Space Pen has the most similar form factor, but I forget how well it actually writes. I have a zebra pocket ballpoint pen that's also compact, but it writes like a ballpoint pen (and with it my handwriting seems to devolve to scribbles despite my best efforts).
lots of (flatbed) scanning today. nearly done digitizing the journals and notes I've done since April. From there, each page gets imported into my digital zettelkasten and assigned a UUID.
After all the flatbed scanning is done, I'll be writing in a way that allows me to use the portable hand scanner "wand" instead. It's a much more portable system, and it's well suited for incremental additions.
Why I'm interested in vocal synthesis
1. It sounds funny. Seriously. Vocal synthesis is one of the raree instances where computer music can have a sense of humor. I find that compelling.
2. It encourages new ways to think about computer generated music. Getting a computer to "sing" often benefits from building new tools.
Retrospective, Angry Ranting
Here's the deal: I guess "machine learning" is a buzzword that apparently gets you money? But this ends up turning things into solutions-in-search-of-problems. You either end up with musicians with superficial understandings of ML saying completely unhelpful things like "what if DALL-E but with music", or data scientists who are too lazy to actually dig into music trying to write programs to do it for them. In either case, nobody is doing anything all that inspiring.
Retrospect
I think I dodged a bullet having my PhD application rejected.
Truth is, I really don't like machine learning. I'm bored to tears by it. And I don't like how it's eating up the music technology space in academia.
During the whole interviewing process, there was no mention of "machine learning", and the "AI" buzzword was used exactly once in the description of the program. Now, I see who they brought on, and what they are doing, and it's full-on machine learning. I would have been miserable.
I teach computers how to sing.