i make beats i’ve never seen and code with a screen reader yelling at me. this site is the center of the chaos: YouTube, TikTok, GitHub, and every app i shipped at 3 AM all orbit this one page. you are now in orbit. no refunds.
One page, with sections for about me, projects, music, writing, the startup chime and links. Use the jump list above to go straight to any section; your settings stay on this page.
This site is one page rather than a set of separate pages. The jump list takes you to each section, and the controls in Settings change how this page looks and sounds; your choices are saved on this device.
Experience (for anyone interested)
hey, i’m Chaotic Carter, aka 2three1y. new name, same chaos. the username stays, it has seniority.
also: i’m blind. yes, for real. i make beats i’ve never seen, i write code with VoiceOver reading every line back at me, and i’ve never seen this website. i hear it’s pretty. if it’s ugly, please lie to me.
that’s why everything i build works with a screen reader from the very first commit. not because it’s trendy. because i’m the first user, and if i can’t use it, it isn’t done. it’s a draft with a domain name.
my main quest: poke at tech until it confesses how it works, then turn it into something people can actually use. some days that’s a browser tool. some days it’s chess you play by ear. some days it’s a quiz that roasts your personality. all shipped. all count. no further questions.
side quest that took over: old computer voices, synths and system sounds. startup chimes, classic speech synths, the tiny beeps old hardware makes. if it beeped in 1998, i want it back on speaker. (you might’ve already heard mine.)
and i love tools that live on your own device and don’t need anybody’s permission to work. local-first, decentralized, actually yours. that’s basically Looscid’s whole personality.
Favorite voices
Classic Alex, the Leopard-era one, before neural voices gave him a midlife crisis
old-school speech synths in general. the more robotic the better. bonus points if it butchers my username
Favorite things
startup chimes and system sounds (i made my own, it’s in the Startup chime section)
synths and vintage soundfonts
retro OS audio that nobody else bothered to save
apps that work offline and keep your stuff on your device, like a normal amount of privacy
Accessibility advocacy (not a paid DLC)
accessibility isn’t a bonus feature you add if there’s time left. it’s the starting line. semantic HTML, focus you can actually see, keyboard for everything, real contrast, structure that makes sense out loud, and no important info hiding behind fiddly extra clicks.
if it only works with a mouse, it’s still in early access. if a div is pretending to be a button, that’s catfishing, and i will find out. VoiceOver snitches. what the hell, div.
Projects (the tabs are multiplying)
everything i’ve shipped. accessible on day one, because i’m the first user and i can’t see the screen either. something broke? ping me. i’m probably already in there.
The headliners
the 11 projects linked on my YouTube channel, in the same order.
Looscid
the main quest. forever in beta, on purpose
my local-first, decentralized web workspace, built around one idea: your stuff is yours. your data and your daily tools keep working right on your own device. no account wall, no “please sign in to continue” jumpscare.
it is not trying to replace iOS or Android. it knows its role. its real job is being comfy with a screen reader: predictable landmarks and headings, short chunks, keyboard-first everything, and status updates that get announced instead of hidden. the goal is a workspace that stays chill even when every single thing is being read out loud. quiet server at 3 AM, not ten people unmuting at once.
every letter you type turns into music. 17 instruments, reverb, delay, a bitcrusher, genre demos so you’re not starting from zero, and Export WAV so your masterpiece survives.
fun fact: my group chat messages go hard as melodies. the group chat did not consent.
learn to make beats with code, for AI agents and humans. it starts with a starter beat that doesn’t suck and upgrades it stage by stage until it slaps. numpy and scipy only.
chess you play completely by sound and keyboard. every move gets said out loud, you type moves like Nf3 or tap squares, and you can play the computer, pass and play, or invite a friend over Nostr with a link. no account needed.
the other 32, grouped. all 43 projects live on this page.
Big builds and experiments
the big swings and the weird ones.
Fork Lab
my forks finally got jobs
i fork a lot of apps. most of them can’t run in a browser. so i built tiny fake-but-working demos of 8 of them: mesh chat on bitchat, posting on Misskey and Feditext, a scripted Kilo Code helper, a TalkBack trainer, the GrapheneOS camera, messages and PDF viewer, Chat UI, and OldTwitter.
all fake data. no servers. nothing to break (i checked, i tried).
a Windows 95-style desktop in your browser, fan-made and totally original. hit Turn on computer for the boot chime, then mess with the Start menu, My Computer, a Notepad that actually saves, and a Recycle Bin with a personality problem.
blue screens so far: zero. knocking on beige plastic.
unhinged quizzes, fake diagnostic verdicts, and internet chaos made to waste your time in the best way possible. your answers don’t get collected. it judges your vibes, it doesn’t harvest your data.
RU a Cheater: step up to a completely unserious verdict machine. it’s a joke generator, not a courtroom. no lawyers, no evidence, just drama.
Persona Evaluator: press a button, get a very official-sounding persona. take it with zero seriousness. your friends won’t. sorry in advance.
a chaotic little model with a personality of its own. its system prompt and the raw training data behind it are open source, so you can see exactly how the chaos was made.
nostalgic chimes, classic speech synths, vintage soundfonts and retro OS audio, rescued before they vanish into the forgotten-downloads-folder in the sky.
open it for one minute. lose an hour recognizing startup sounds. happens every time.
a tribute to the legendary, uncompressed Leopard-era Alex voice, from before neural models gave him a midlife crisis, weird pitch, and a toxic relationship with compression. Alex, sounding like Alex again.
GitHub Pages is off for this one, so it’s source only for now.
small apps that do one job and don’t waste your time.
Earshot
it reads to you and never gets tired
an offline talking assistant in your browser. paste text, open a file or a link, and it reads it out loud with natural neural voices, or hands it straight to VoiceOver. talk to it to pause, skip, speed up, or ask what the text even says.
voices only download when you press Load voices, so it’s not eating your data behind your back. unlike some apps. you know who you are.
fast reactions and accessible toolbar shortcuts, so you’re not digging through menus like an archaeologist looking for the sacred Like button. hit the reaction you want while the moment is still happening.
convert files without 47 sketchy ads, fake download countdowns, or a “Click Here” banner trying to install a toolbar on your grandma’s laptop. it all happens right in your browser. your files stay on your device and never get sent anywhere.
a two-panel code playground with a language dropdown and live output. run JavaScript, preview HTML and SVG, and draw on a canvas. press Run or Ctrl/Cmd + Enter. Python is in the list but can’t run yet. despite the name, it runs or renders code, it doesn’t translate between languages.
press play and your Nostr feed comes in as sound: an earcon for every note, reply, repost, reaction, zap and mention, then the note read out loud (or by VoiceOver). safe curated relays by default, plus an optional station for people you follow.
accessible live Nostr audio rooms and streams. earcons tell you when a hand goes up, a mic turns on, or a room ends, and the live chat gets read out loud.
it’s listen-only and opens Nests rooms in their own app. basically a voice channel without the guy eating chips on open mic.
a serverless Morse code chat. share a room link, tap the big pad, and your friends hear and feel every dot and dash. it has a learn mode and a braille view.
built only from repos i’ve starred, every one credited. the stars finally paid off.
the bot in the Looscid Discord that celebrates each language’s holidays in that language’s own channels. Eid in the Arabic chat, Diwali in the Hindi chat, Día de Muertos in the Spanish chat, and nobody gets spammed with everyone else’s.
the code is public so you can read it and learn from it. the bot itself only works in Looscid, and it leaves any other server.
i can’t see a waveform. don’t need to. i hear the 808 and i just know. digicore, jerk, hyperpop, whatever the 808 demands that day, and the beat switch is always the best part. that’s the law.
latest beat: still in the oven. new Chaotic Carter beats drop on YouTube first, and they’re free. turn your volume down first. or don’t, i’m not your mom.
heads up: this part is mostly bass. like, a concerning amount of bass. the numbers below are my rough measurements of each track, not a science fair project.
right now: whatever shuffle pulls from my 203-song playlist. expect hyperpop, digital rap and jersey club bounce, mostly under 3 minutes, anywhere from a laid-back 80-ish BPM to a 170-ish BPM sprint, with bass so loud the loudness meter filed a complaint.
Playlists
my song playlist: 200+ songs, mostly SUPXR, waera and pxiqzes. shuffle at your own risk. it slaps, every time.
YOU DECIDE by SUPXR: hyperpop with a jersey club bounce at about 129 BPM. the only thing you actually get to decide is how hard you nod.
wussup? by shteppi and SUPXR: about 123 BPM and roughly 69% bass. it’s the song version of someone typing “wussup?” and joining your voice channel uninvited.
concord. by shteppi: waera on the beat, around 136 BPM, mastered so loud (about −8 LUFS) it doesn’t knock, it kicks the door down and asks why you’re muted.
ronald. by shteppi: a lazy bounce around 83 BPM with about three-quarters of the energy in the bass. like the server member named Ronald who never talks but everyone loves anyway.
SCRUPLES by yit and Isaac Horner: about −5.7 LUFS, the loudest track on this whole site by a mile. mastered like it just bought Nitro and needs you to know.
press start. by pxiqzes featuring mi.ntyyy: about 150 BPM played half-time with 72% bass. chill game lobby energy, right up until the boss music hits.
flower by waera: fast (around 172 BPM) and the bassiest one here at about 89%. calling this “flower” is the biggest lie in the server, respectfully.
What’s a Nintendo? by Lilithzplug: a bouncy 129-ish BPM, about two-thirds bass, with a title built to start a 40-message argument in the gaming channel.
Articles (Writing, because brevity failed)
for when a thought needs more than one breath, or more than six messages in a row. latest:
I love the Acapela voices: Sharon and Sharona. Alex too, the classic Leopard voice I’ll keep forever. I love how they read, and how they hold up when sped up. That’s it.
i made my own startup chime. it stays off unless you turn it on. turn it on and it plays once when the page opens, or on your first tap, click or key press if your browser makes sites wait. it’s soft, about two seconds, made live in your browser, and quiet enough to sit under VoiceOver. your pick is remembered on this device. if your iPhone is on silent, it might stay quiet.
Contact (the internet has many doors)
pick a platform, any platform. i’m on all of them and asleep on none. they all point back here anyway.
good news first: nothing on this site flashes or moves, and it stays within the three-flash safety limit. this is the one section where i keep a straight face. mostly.
More details (the actually important part)
what got checked: every animation, transition and script on this page, plus a frame-by-frame screen recording measured against the WCAG 2.3.1 limits (no more than three flashes in any second, red flashes included).
nothing on this site moves or blinks on its own. not even a little. the big color blocks and rings up top are painted on and stay perfectly still.
your device’s Reduce Motion setting is respected: if anything that moves ever gets added, it stays still while that setting is on. Calm mode in Settings is on by default too. your settings outrank mine, as they should.
sound: the only sound here is my startup chime, and it’s off unless you switch it on (the Startup chime button in the Startup chime section, or in Settings). when it’s on, it’s soft, about two seconds long, and plays once when the page opens (or on your first tap, click or key press if your browser makes sites wait). nothing else on the site makes sound. ironic, considering the playlist.
honest limits: this was tested with automated measurement, not a clinical tool like PEAT, and not on every device out there.
if anything here ever flickers, that’s a real bug, not a feature and not a vibe. please report it on GitHub and it jumps to the top of my to-do list, ahead of sleep.
Settings (the knobs are labeled)
Set the page up your way. The controls save as soon as you change them; no Save button is hiding in the bushes.