#Steno Reimagined for a CC2 Layout

196 messages · Page 1 of 1 (latest)

molten geyser
#

Yo not sure if anyone’s thought of this before, but here we go. FYI I’m a total noob and just learned all this steno & CharaChorder stuff for the first time.

Why not just make steno on a charachorder, but cut down on the key-combos to make it possible+easier. Like instead of having TPKW = g- (left bank g), just have a dedicated g key on your left hand, and have a dedicated g on your right hand for -g. Then have AEOU on the thumbs just like on the stenograph. Here’s a list of sounds that physically can’t be typed simultaneously on each bank of a stenograph (according to ChatGPT) so that we have a good list of candidate keys that could pair up on each finger:

Left bank —
Sound
Conflicts (share a key)
B
F, G, J, M, N, P, W, Y
D
F, G, J, K, N, T, Y
F
B, D, G, M, N, P, T
G
B, D, F, J, K, M, N, P, T, W, Y
H
L, M, N
J
B, D, G, K, L, R, S, V, W, Y, Z
K
D, G, J, Y
L
H, J, M, N, R, V, Y
M
B, F, G, H, L, N, P
N
B, D, F, G, H, L, M, P, T
P
B, F, G, M, N
R
J, L, V, Y
S
J, V, Z
T
D, F, G, N
V
J, L, R, S, Y, Z
W
B, G, J, Y
Y
B, D, G, J, K, L, R, V, W
Z
J, S, V

Right bank —
Sound
Conflicts (share a key)
B
J, N
D
—
F
—
G
J
J
B, G, L, M, N, P
L
J, M
M
J, L, N, P
N
B, J, M, P
P
J, M, N
R
—
S
—
T
—
Z
—

So for example, since you may have to be able to press R & D at the same time on the right bank (maybe in something like “heard” you’d needa type -rd), you wouldn’t wanna put -R & -D on the same finger. Also for example, for sounds like G-, since G- is formed by simultaneously pressing “t- p- k- w-” on a stenograph, B- could be a good candidate to be on the same finger as G-, as B- is formed by pressing “p- w-”, so you literally can’t simultaneously press G- & B- on a stenograph. This could be stating the obvious, but on a single 5-way switch, you can’t press any of those 5 keys simultaneously, so we have to be careful which keys we put where on the charachorder.

#

As far as making the layout goes, you’d have some code optimize a regular 26 key-layout with some list of restrictions to prevent putting incompatible keys on the same key. You could even allow it to place the vowels wherever, and just have a second set of vowels on the thumbs for steno purposes if there are enough keys. So obviously, this layout would probably be worse than the default CharaChorder layout in terms of chentry speed since it has some restrictions placed on it. The hope would be that this would only cost 10-20 wpm of chentry.

#

So then, let’s say from that optimized layout, a G key happened to land on the right hand. That would mean you would need to place a secondary G key on the left hand somewhere strategically that doesn’t cause any conflicts. There’s probably a lot of possibilities here too, so you’d probably want to use some code to optimize the second step too. Also, the secondary G key wouldn’t actually produce a G when tapped on its own– it would produce a special character, or maybe it would be an arrow key, or maybe it would be a modifier (not sure whether CCOS let’s you chord with modifiers or arrow keys or not but that would be helpful) (I don’t have a CC2 FYI). Like maybe if you were practicing memorizing the layout, you’d use a separate practice layout that would have the left bank/hand keys produce capital letters and the right bank/hand keys produce lower case letters idk. Also, I noticed that on the default charachorder layout, there are no letters on the pinkies and the pinkies are mostly used for modifiers and other special keys. I’m not sure if the reasoning is that pinkies are just like considered “weak” homerow keys or that people like to not have any other keys near the shift key or maybe for rolling purposes. My opinion is that if I had, say, both “h” and shift on my left pinky, then I would just have to use right shift to capitalize h’s. Also, I wouldn’t mind putting common shortcut letters (e.g. x,c,v,w,f,t,tab,a,y,z) on the same finger as the finger that presses ctrl, because I could just use AHK to remap certain shortcuts, or I could just use right ctrl for certain shortcuts.

#

Feel free to skip this paragraph: Also just a quick rant on the default layout– it seems like a bunch of letters are clustered on the thumbs to compensate for the lack of letters on the pinkies. Having a bunch of keys on 1 switch can allow for a lot of nice rolls for bigrams, but theoretically rolling only helps with pressing keys in the right order (aka accuracy), not necessarily speed. Theoretically, distributing the keys across more fingers would build speed by reducing the likelihood of 2 consecutive letters needing to be pressed. Theoretically, it’s faster to hit two separate keys consecutively across two fingers rather than rolling because there will always be a small delay between keys when you roll them, but you can arbitrarily reduce the delay between two consecutive keys strokes on separate fingers (it just requires you to have good timing, which humans probably don’t have, at least consistently). It’s probably the case that in practice, rolling is probably consistently more or less always faster for humans to accurately press 2 consecutive separate keys. So yeah with this layout, since there are so many keys, it’s probably inevitable that we’d have to assign some letters to pinkies. Maybe we’d only assign secondary keys to them or maybe we’d assign some primary keys to them but try not to reduce the chentry WPM too much in doing so.

#

I also think Conjugation/English Grammar Theory is cool. I read some of that thread, but not the whole thing but here are my thoughts. It would be cool if we could somehow squeeze in a couple directions for conjugation in the layout. Maybe dedicating 8 directions to conjugation would be too much and would mess with the space limitations (there’s only 4*12 =48 good homerow keys) because with this layout, we are already gonna be tight on keys. It’s my understanding that steno already has some built-in endings-appending methods, but it doesn't work for every word and sometimes you just have to tap an extra chord after the word to conjugate it (it doesn’t surprise me that steno has such a steep learning curve). With conjugation theory, you could save a tap by consistently having the same keys always be available to conjugate a word. Also, of the 12 english tenses, only like 5 of them are like actually used often (the others are used like <1% of the time according to ChatGPT). So maybe some key combos that would originally be for the lesser common tenses like future perfect continuous (e.g. will have been doing something), maybe those key combos could be repurposed for some other more common conjugations, that way we can cut down on the required direction/key count for conjugation (so maybe instead of 8 directions, only 4ish).

#

Anyway, I think by cutting down the key combos people have to memorize and by having a consistent set of conjugation keys, a layout like this could be easier to learn+could be faster than a stenograph. Also, we could just take a Plover library and do some tweaking to it somehow to accommodate the layout because it’s really not that much different from regular steno, so I imagine you wouldn’t have to build libraries and brief libraries from scratch. I think it’s best to build a brief library in advance because it sounds like a nightmare creating and tweaking briefs on the fly and having to unlearn muscle memory (seeing from that 1 thread where someone talks about how they struggle with that). Easier said than done but hopefully we could just tweak a Plover library. Please let me know your thoughts. I had a couple other ideas that involved having only 1 of every key and using disambiguation for anagrams, but this theory seemed to be the best because I feel like the disambiguation could slow things down a lot.

west marten
#

@wide raft (I’m at school rn but I figure you could provide some commentary on this)

wide raft
#

Yeah I already made a steno layout
https://discord.com/channels/861730583092658206/1394694396980756731

on a single 5-way switch, you can't press any of those 5 keys simultaneously
Yeah, so on Michela (which is steno designed to work on a piano) There's the same restriction
🎹, a finger cannot press two keys at once

Which means Michela steno fits on a CharaChorder

image-9.png
#

The way Michela works, is for each consonant sound, you memorise a somewhat random combination, SP = "ch"
I say somewhat, because the more common the letter, the fewer keys you want it to use

image-10.png
wide raft
#

I did not do that, because Michela steno already works, and I do not own a CharaChorder, so it was fun enough just to get a working theory, but this would be easy to implement as it doesn't have knock on affects (other than freeing up more chords)

sage trenchBOT
#

Here's the best 2/15 entries for 'character' in Michela theory


CPa/FCNipc/FPencf → character
           CP │ ch pronounced k
            a │ short a
            / ┐ 
          FCN │ r
            i │ EU vowel spelt a
           pc │ c pronounced k
            / ┐ 
           FP │ t
            e │ e
          ncf │ r``````Ansi

CPancf/CPRIUencf → character```Too big for Discord, stopped annotating
wide raft
#

It'd be the same amount of strokes, because theory basically hasn't changed

but fewer keys, and easier to read
KRak/Ter -> "character"

wide raft
molten geyser
shut irisBOT
#

GG @molten geyser, you just advanced to level 1!

wide raft
#

You can use the plenty of unused keys

All entries start with the left hand, so if you need more keys, the right hand is always free

Or you can forgo chentry, and orthospell 🦐 (fingerspelling, but can do multiple letters at a time)

#

I'm biased, because I do all my coding on a stenoboard, I no longer need chentry
But that original layout has many keys unused

molten geyser
wide raft
#

WSI has space for 2,621,440 unique combinations

Michela has space for 186,624

Not sure how to work out how many combinations CC has space for, because without a device, idk if it's even possible to hit combinations like
pinkie east + ring north

#

But Michela fits into CC with many spare keys. These spare keys could replace chords, like replace FCN -> 'r' with R -> 'r'

Then maybe come back and map FCN to something now that's free

molten geyser
wide raft
#

It's used in Italian courts, it's the default theory there

#

It tends to have more strokes than WSI, but much lighter strokes, since you're never expected to hit multiple keys with one finger

#

There's no logic why the 27 combinations are mapped as they are, I might come back to this, use the extra keys CC offers, and make sure the most common letters use the fewest keys

But I'm working on a steno theory that'll fit on a 🎮 first lol

Screenshot_20250807_152036_com_discord_MainActivity.jpg
molten geyser
molten geyser
#

New idea.

It's probs highly impractical to fit 2 steno alphabets (1 on each side) to recreate steno on a CC2. But then i got to thinking. The reason you can fit so many letters on a stenograph is bc 1 finger has so many combos. Like for a 2x2 array, u can do 4 keys at once (+1), 1 key at a time 4 ways (+4), or 2 keys 4 different ways (+4) for a total of (9) combos for 1 finger. (technically it's probably more like ~7 for the middle and ring fingers & 8 for the pinkies and pointers because of overlap, so a total of ~30 combos per hand). Meanwhile on the CC2 each finger can only hit 4 directions with the exception of the thumbs.

But what if we tried to get the middle & ring fingers to actuate multiple keys at once? We could take some molding puddy and build up the 4 small keys until they touched our ring and middle fingers. Theoretically 1 finger would have the power to actuate 4 directions for the 1st switch, 4 for the 2nd, & 4 combos (both north, E, S, W) for a total of 12 directions for the rings, middles, & thumbs. So 3*12 + 4 from the pointer & u get 40 total directions per hand not counting the pinkies. All of a sudden, steno starts sounding a lot more viable on a CC2.

Not sure if the Charachorder team had this idea before. Maybe they thought it was too impractical. It's probably slow to switch b/w combos and individual directions within the same finger. Also, many multi-finger chords are hard to perform, which reduces the # of possible chords a lot (e.g. it's hard to have to fingers squeeze inwards towards each other), so there's still may not be enough chords for steno.

Maybe it's just a feature to request for the future CC3? It would be cool if the pointer and pinky also had 2nd switches. I mean it would be cool if each finger had 3 switches. Maybe there's a hope 1 day Charachorders would fit inside laptops, and maybe to make things viable, some of the 2nd and 3rd switches would have to stick up a lot, which could complicate the laptop idea.

#

I had ChatGPT Thinking do some math for me for the number of possible viable chords on a CC2 (idk if this is a known number). Here are the restrictions I had it use (idk if someone’s made a comprehensive list out there):

🟥 No pinky usage
🟥 No 3-finger chords among {index, middle, ring}: at most 2 of those 3 fingers may be active in a stroke.
🟥 No “Down/in” direction anywhere (only N/E/S/W).
only 12 directions allowed for thumb, middle, ring
🟥 Index toward Middle requires Middle idle
🟥 Ring toward Middle only allowed if Middle matches or is idle
🟥 Pinch-the-middle (both Index & Ring toward middle) forbidden

The total without allowing middle and ring to hit 2 switches at once apparently is 474,720 & the total for allowing 2 switches at once for middle & ring is 8,708,400. The upper bound for a stenograph is 8,388,607, but in practice it’s more like 2,621,440 for WSI steno according to Harri (how’d you get this figure btw?). But yeah maybe someone could check these figures.
Also, I didn’t realize this, but I guess there’s like a steno order that’s important for occasionally squeezing multiple syllables into a single chord: #STKPWHRAO*EUFRPBLGTSDZ . This order would likely get messed up on a charachorder layout. I mean like we could still use this order, but it would make it harder to remember the order cus like on a stenograph, the order is in an easy-to-remember zig zag pattern. So yeah basically, I imagine we wouldn’t have to completely reinvent the wheel in terms of being able to use existing steno chording libraries, but there’d probably have to be some tweaking done and a new order made. But it could theoretically be done programatically i think. I think the steno order is just kinda arbitrary anyway and it seems like a consequence of optimizing for other things– like i think whoever made the steno layout didnt prioritize the combining of 2 syllables into 1 stroke via the order thingy as much as they optimized it for other stuff. Maybe I’m wrong.

#

Lemme know y'all's thoughts

wide raft
wide raft
#

but if you don't want to do that, I suggest you learn the 16 vowel chords that all English steno theories use
then you look at the most common English starters, make sure they're accessible to the left of the vowel chords, and look at the most common English enders, make sure they're at the end of the vowel chords

image.png
#

It's more important you can write "str" than "x", so don't think you only need the alphabet

molten geyser
# wide raft You could start with this already proof of concept steno layout that fits on a C...

I wanna try to go for the endgame fastest steno theories, not Michela.
I like your idea of taking the most common combinations and assigning them to their own dedicated directions/combo of directions, and then using the freed up fingers for other things.
It seems ever more possible with my new idea. Do you think my idea is viable or do you think pressing multiple switches with 1 finger at once is too finicky?

btw are these graphs for Michela or regular steno or something else? are they like actual letters or like phenomes?
Where did you get these graphs? id like to see more of them

also hows the steno on a controller 🎮 coming?

wide raft
#

I don't own a CC so I can't say
I would assume pressing 3 keys up is easier than 1 finicky key

#

These graphs are just analysing English, I made it by analysing the CMU corpus

molten geyser
#

so like letters not phenomes?

#

like what is meant by HH

wide raft
#

I'm making steno controller theory part of my uni work, I'm splitting it into 2 parts

  1. using an evolution algorithm on the WSI (the standard for English) layout to see if it discovers a layout that can come up with more one-strokes without increasing the conflict ratio

And it has!!! look, fitness score goes up!

image.png
#
  1. using this evolution algorithm, I hope to evolve a layout for controller steno
#

But first I gotta write my paper for the first one that's due next week

molten geyser
#

whats a conflict ratio

wide raft
#

So
TK -> d
PW -> b
TKPW -> g
These rules are pretty standard, and create loads of mappings that don't conflict, however there is one conflict
TKPWAEUT -> gate/debate

So I'm happy with the level of conflict WSI has, cause it's really not that bad, but I don't want it any worse

molten geyser
#

oh i didnt even know steno had conflicts. i thought they strategically designed it to make conflicts impossible

#

So what is HH on that 1st graph btw

wide raft
molten geyser
#

so they are phenomes not letters?

#

sry im bad at steno jargon

wide raft
wide raft
molten geyser
wide raft
#

And in steno, consonants are sounds
Vowels are annoying cause they pay attention to both spelling AND sounds, because otherwise homophones would have the same mappings
So I'm not gunna attempt to change vowels

molten geyser
wide raft
wide raft
molten geyser
#

ok

#

so to make conflicts impossible, do they typically have to sacrifice some efficiency?

#

seems like the steno layout was made back in 1910 before AI could help optimize the layout

shut irisBOT
#

GG @molten geyser, you just advanced to level 3!

wide raft
#

Yeah, so for orthospelling you can't drop vowels, because dropping vowels introduces ambiguity and sometimes conflict

dropping unstressed vowels is very obvious for stenographers
BREEV -> bereave
BREETH -> breathe

#

But with orthospelling, it's always 1:1 matching, so you'd have to have a dedicated 'be' chord instead of
PW -> b AND be AND bu AND etc

wide raft
molten geyser
#

oh nice

wide raft
#

The WSI layout is very good, the creator is so much smarter than I am, his choices are so good
But hopefully with AI (evolutionary algorithms specifically) I hope to create a better layout

#

Other than his patent I cannot find what guided his layout

molten geyser
#

is there a steno discord where people talk about steno theories

molten geyser
wide raft
molten geyser
#

thnx

wide raft
molten geyser
#

when do u finish ur AI degree. do u plan on optimizing a theory for a charachord device

#

im debating whether i try to make my own quickly or just give up and settle until someone carries and makes an endgame stenoesque theory

wide raft
#

Without a CharaChorder myself I don't know which combinations are easy/hard to hit

On a regular steno keyboard, I know that it's hard to hit -TZ and -SD, so I exclude all combinations that have that combo inside it from being analysed

Maybe on a CharaChorder ring finger east + middle finger north is hard to hit at speed? So I would exclude that combo

image.png
molten geyser
#

Some dude timed like 2000 different key combos. I asked him for his code but he never responded

#

He wanted to make a more accurate cost function as opposed to just ranking your favorite directions

#

Dude wrote like a 40 page paper on it

wide raft
#

But as there is no documentation for what chord combos are compatible with what other chord combos, I can't do much.

Which is why my first attempt with the Michela layout only has North and South key directions, as I suppose they're always compatible to combine

molten geyser
#

i have a CC2. the restrictions i listed seem somewhat comprehensive

molten geyser
wide raft
# molten geyser I had ChatGPT Thinking do some math for me for the number of possible viable cho...

the steno order is just kinda arbitrary anyway and it seems like a consequence of optimizing for other things– like i think whoever made the steno layout didnt prioritize the combining of 2 syllables into 1 stroke via the order thingy as much as they optimized it for other stuff.

What I'm finding, (and trying to undo) is that it's optimised for the average English syllable
where -R commonly follows a vowel

In the average word however, -R can come after a consonant, usually ends a word
"father" I would like to write F A TH R, but in WSI, there's only one -R after the vowel, and it comes before -TH, so "father" has to be two-stroked

molten geyser
#

oof

wide raft
#

One vowel per word is such a waste when there's so many words with an unstressed 'a' at the start
again, about, attain etc
and so many words with a -y at the end
funny, happy, trolley etc

So I agree I think the WSI layout is not optimised for multi-syllabic words

shut irisBOT
#

GG @wide raft, you just advanced to level 19!

molten geyser
wide raft
#

And I can't find Mr. Ward Stone Ireland's goals for his layout!! I'm confident he optimised it for the average syllable but I can't find that written anywhere!

molten geyser
#

from that thread

"

I think it's something that's really hard to quantify
I assume you mean speed and comfort
Generally a horizontal pinch is really uncomfortable to impossible
Diagonals are possible but unreliable
left/right dual thumb presses are possible and viable imo
(aka gm kw hDUP fb)
Lower thumb down is harder
Thumb/other finger pinches are very comfy
vr/ve is hard but doable
or/an is comfy
ny/yj is next to impossible
I think that's fairly comprehensive?
Using more than one finger seems like it should be harder but really isn't, it's more of a function of which parts of the combo fall into which difficulty categories
uoecdtns is on paper a hard chord because it's long but it's an easy motion so it's much much easier than jy

"

wide raft
#

I could create a bunch of cost functions
The comfier the layout, the fewer combinations there are to play with, but as I don't have a CharaChorder myself, I wouldn't be able to ground myself with "these chords look impossible!" or "there's so many unused chords!"

molten geyser
#

do u plan on getting one?

wide raft
#

My code is openly available for someone else to play around with, I could create a few presets so someone with a CharaChorder could edit it from a given prototype

#

Not really, the strength of CharaChorder is that you're eased into chording, whereas steno requires you chord everything.
But since I've already learnt a steno theory to the point I can chord a word I've never seen reliably, there's no benefit to learning CharaChorder

molten geyser
#

ergonomics

#

and it'll probably surpass steno 1 day with the proper theories

#

in terms of speed

wide raft
#

Steno is more ergonomic, as the force comes from the big bicep muscles (so it's your big elbow doing the force to actuate all the keys)
It's unlike Qwerty where you press one key at a time, in steno, your fingers position themselves, and then your arm slaps down
Your fingers muscles (green and brown) only have to deal with the force of moving themselves, not even the force of actuating the keys

image.png image.png
wide raft
molten geyser
#

steno has phonetic mnemonics to remember chords so u dont have to memorize infinite briefs

wide raft
#

Most steno theories are phonetic yes

#

So it's because chord theories are spelling based?

molten geyser
#

i guess

wide raft
#

I consider chord theories to be a type of steno theory, that has to make space for chentry

#

Whereas in traditional steno theories, all combinations are devoted to steno

molten geyser
#

dont u have to move your fingers around tho on a stenoboard where as on a charachorder they just kinda stay in place mostly and wiggle?

#

the act of positioning ur fingers for steno uses finger muscles

wide raft
#

Yeah, your finger muscles only have to position themselves, they don't have to actuate the keys

#

Yes, you move your finger up to 1 key length

#

On a Qwerty keyboard, you move 2 key lengths
R-> F -> V is double the distance steno ever does

molten geyser
#

i feel like the distance u have to move for charachorder is less so maybe one day itll be faster

shut irisBOT
#

GG @molten geyser, you just advanced to level 4!

wide raft
#

In Qwerty I can believe a limiting factor is how fast your fingers can physically move, but for steno hesitation is the limiting factor until you reach speeds of at least 200

molten geyser
#

is orthospelling efficient for coding? u say u use ur stenoboard for everything, but do u ever run it scenarios where u wish u had chentry?

wide raft
#

Ah, this is where CharaChorder might have me beat

I like coding with steno, because everything is right under my fingers already, I don't have to move my hand to reach }, because with steno, everything is maximum 1 key away

With CharaChorder, you also don't have to leave the joystick, so you get the benefit of not having to reach for stuff like } too

wide raft
#

But steno is no faster than Qwerty for coding. Steno coding is just for ergonomics

molten geyser
#

the other i like is the ability to roll. Think of a mouse wheel-- it allows u to actuate a key over and over again very quickly- until u reach the end of the scroll wheel. The switches on a charachorder are like scroll wheels with no end-- u can just keep going round and round.

#

i use excel a lot so im planning to map multiple of each arrow key to CC2 switches so i can be lightning fast around a spreadsheet

#

idk

wide raft
#

Steno has a tap and hold feature, then it spams that entry

molten geyser
#

yeah but u cant stop and start as quickly

wide raft
#

What do you mean?

#

Quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly quickly

#

I hold my entry for "quickly"

#

tap and hold

#

it's good for navigation, where you tap and hold a direction

#

or ctrl+backspace

molten geyser
#

dont u have to press and hold for a second to wait for the repetion to start

#

plus if u wanna stop it on a dime i feel like its less precise than being in control of every key press

wide raft
#

To stop, I just lift my hands

#

but yes, to start, I wait a few ms

molten geyser
#

yea but its going at 30 times a second

#

my theory is that if i had 3 CC2 switches going round and round each NSEW are all mapped to down arrow key, and i had a specific place i wanted to go to on a spreadsheet, i feel like I could swirl all 3 at once for a combined ~20 times a second, and be able to more reliably stop on a dime on a certain cell in a spreadsheet cus im in control of every press. idk just a random niche thought i had

#

also when ur starting and stopping a lot and making micro adjustments in a spreedsheet, those couple of initial ms of delay are a lot

#

only rly important if ur a spreadsheet guy i suppose

wide raft
#

I use -S and -T to increase the outputs

STPH-G → "{#Right}{^}"
STPH-GS → "{#Right}{^}{#Right}{^}"
STPH-GT → "{#Right}{^}{#Right}{^}{#Right}{^}"
STPH-GTS → "{#Right}{^}{#Right}{^}{#Right}{^}{#Right}{^}"

so I can do 4 rights all in one go

#

Any more than that and I do the tap and hold

molten geyser
#

yeah ive thought of that and im probably gonna have to do something like that for really far moves but once u get in the ballpark of where u wanna go ud wanna drop the multiplier and still be able to go really quickly and precisely with the 1 at a time arrow keys. my idea could be a bad idea idk i needa try it

#

also using an eyetracker somehow might be the best solution for me anyway idk

molten geyser
#

Also idk how large the span of "fairly optimized" steno layouts are. If that space is large, we are bound to find one that also happens to be good for chentry. Also, given the new access to more directions and the fact that we already are gonna basically have 2 of every letter for steno purposes, ppl could also use the extra letters to boost their chentry idk. or maybe the chentry turns up being really bad and so maybe the extra letters help off set the badness.

maybe we optimize for steno and find that the most optimal layout can sustain speeds of 350 WPM but that ends up messing up chentry to only being 50 WPM. but maybe we optimize for chentry first and find decent steno setups with that give 330 WPM but 90 WPM chentry. we'd have to prioritize what we prefer. i say we prioritize chording

molten geyser
#

@wide raft

Aren’t stenograph switches like extremely light to reduce strain and improve speed— like 20-30g compared to like 60g for a regular keyboard?

Seems like it improves ergonomics while typing, but maybe it hurts ergonomics at rest when ur not trying to type. Like I imagine a good portion of your time in front of a screen is spent reading stuff or thinking in between bouts of typing. So during that time, it seems like you have to constantly hold back/fight gravity to make sure you don’t accidentally fat finger something? Maybe u have a script prevent the chord where all the homerow keys are down at once, and u just let ur fingers always fall? Maybe u can get a split stenograph and mount them a lil diagonally/tilted to minimize some of the weight of ur fingers and also cus some tilt is more ergonomic. I mean u can always take ur hands off but that sucks.

The down press for a CC2 is really heavy, and u would never hit it just by having ur fingers rest on it. From my experience, it is possible to have the weight of ur fingers accidentally push the north direction, but I imagine it’s not as bad as a stenograph and can be mitigated some by mounting or like using a chairachorder setup.

Lemme know ur thoughts.

wide raft
#

Typing Qwerty with a steno keyboard is ass because the keys are too light
With Qwerty you're doing letter by letter so it's easy to fat-finger

With steno, it's harder to fat-finger because you position yourself above the keys, and then splat splat splat
So it means we can have ridiculously light keys, my keys are 12g, I would not be able to Qwerty type with these

The heaviest switches for steno are 45g, and it's highly recommended to switch them out.

Professionals don't with do that "between bouts of typing." because they're transcribing what everyone in the courtroom is saying

image.png
#

Which is why no one cares about "30 second sprints" when it comes to steno, because that's not how it's used

whereas yes, for regular keyboard usage, like programming, you think for a few seconds, and then write for a few seconds

molten geyser
#

But like I’m saying for everyday pc usage

#

Yeah

#

Do u use a stenograph for ur day-to-day?

wide raft
#

Yeah, I use it for day-to-day, in fact last year I didn't even bring a Qwerty keyboard to Uni with me

#

TKAEU/TKA*EU → "day-to-day"

#

If I'm gonna do coding, I stuff my keyboard in my pocket and plug it into the Uni computers

molten geyser
#

Does holding ur fingers back from falling cause u fatigue

wide raft
#

No, I position the board really low down, so my palms are resting on my thighs

image.png
#

I'm not holding my hands up, they're dangling, basically resting on my thighs

#

Also I switch up my position so much

#

Which I do on a Qwerty keyboard too, I'm just restless like that, but on this stenoboard it's that much nicer because I can stick it on a magic arm and reposition it

#

I got the magic arm idea from this server! Thank you CharaChorder server, most stenographers use a big tripod

#

The ideal keyboard would be a split bluetooth stenoboard with two different magic arms, but it's really expensive, so next best I think is a split controller (about 15$) so I'll buy one of those to reward myself when I complete my controller steno theory

#

Ideal layout is the Chair-a-Chorder but since I have gone through the effort of learning steno, I'd use a split stenoboard instead

#

Here's a pic of a professional
Very low

1QXR1g2wGMFGp_uwreOqrqQ.png
molten geyser
#

Gotcha. but even when you rest your palms on your thighs, cant the weight of your fingers still accidentally hit those feather-light keys? A palm rest ≠ a finger rest?

molten geyser
wide raft
#

I will not change the theory I already have on my stenoboard

I've met stenographers with awful theories (particularly the older professionals as they learnt their theory back in the days of paper tape, where after court they would rewrite everything into English) and because they know their dictionary like the back of their hand, they're just as fast as any other theory

A good theory means you don't have to memorise as many briefs and conflicts, but I've already done that, so it's too late to switch.

But this first project has shown me that my fitness function leads to a better layout than I could come up with on my own.
So I will reuse this fitness function for when I evolve a controller layout

#

Words that WSI struggles with like "piano" are three-stroke, but rare enough I've already got them memorised

#

It's gonna be weird relearning steno on a controller, reaching for keys that aren't there

#

But that's project 2

#

This is why I think the top speed for steno is the same at the top speed for CharaChorder, once you know your dictionary, it's just how fast you can recall it

Which is why I think it's a shame CharaChorder doesn't have a good theory, it makes the learning of 100,000 words so much harder. But once they're muscle memory, theory doesn't really matter

molten geyser
wide raft
#

Chording is faster than chentry
So at the upper speeds you're chording everything
If CharaChorder had a good theory, it would be equivalent to steno, cause it would be chording everything, it would be steno

However, CharaChorder has chentry readily accessible as a bonus

#

I think there's a video out there of Riley transcribing a video, and he's chording everything, that's basically steno

#

So I would like to bring steno to CharaChorder, however it would be nice to preserve the chentry ability, not for speed, but it's nice to have

molten geyser
molten geyser
shut irisBOT
#

GG @molten geyser, you just advanced to level 5!

wide raft
# molten geyser i feel like since the travel distance is shorter and the fact that it has room f...

Up till about 150-200WPM the limit to how fast you can type is how fast you can recall the outline

after that speed, the limit then becomes how fast your fingers have to move to physically keep up with what you're thinking

Magnum suggests to get past this, you should have more information per stroke, stuff like TPWRO*FT → "in front of the" all in one go. Then your limit is how fast you can recall these briefs instead of how fast you can type "in" + "front" + "of" + "the"

you're suggesting that to get past this, you need to reduce the distance between each key, and I agree that would make it easier, but eventually you'll run into the limit once again, and you'll be forced to memorise more phrasing/briefs

image.png
#

You have to be going at literally record breaking speeds before the distance you're moving your fingers starts to matter.

#

I talk to professionals who have passed their 225wpm test, and they're still writing things out the long way

wide raft
#

It doesn't take any energy to rest on your palms with your fingers dangling

#

I do not think Mark Kislingbury (the fastest stenographer)'s hands are limiting his top speed, I think it's how fast he can recall his briefs, and how fast he can decipher speech

molten geyser
#

new idea (could be bad fyi):

ok what if we swap in bigger thumb switches for the 3rd thumbs switches to make them more accessible, & allowed the thumbs to make even more combos. i did the math, and each thumb would be able to probably press 24 combinations comfortably. So intead of 40 of combos per hand, that becomes 52. So theoretically, we could maybe, just maybe, fit another set of vowels and a 3rd bank on the CC2, allowing for much more 2 syllable words to have easy 1 stroke chords that don't require memorizing extra briefs. Like ik there are a couple tricks for squeezing 2 syllable words into 1 stroke like using the steno order and dropping unstressed vowels and hoping there's no contradictions or occasionally allowing for swapped letters, but it seems like these only work on a small percentage of 2 syllable words. According to chatgpt, 1 syllable words make up 30% of english words, 2 syllable words make up 36%, 3 syllables make up 22%, and 4 syllable words make up 8%.

According to chatgpt u can have up to 3 consant sounds combined in an onset and 4 for a coda (rarely 5) e.g. one word i can think of that has a lot is like strength has "str" and "ngth". Just because we have a lot of combos, doesnt guarantee we can squeeze in another vowel and consonant bank in cus we still have to have enough fingers at the end of the day to execute it. So like if we had a CCCVCCCCVVCCC type word we'd be cooked. So what we'd probably have to do to increase the possible number of 2 syllable words we can make is we'd need to have dedicated directions for common key combos e.g. the ones from those graphs u showed earlier. Idk how we'd go about differentiating which consonants go in which bank but maybe the fingers that r freed up from 1-direction=multiple-letters combos count toward the 3rd consonant bank.

The dream would b to have 3 switches for every finger, but a CC device like that may never come out & i imagine 2 switches for middle and ring will be difficult enough.

#

also like the thumbs are unique in that u can "jut an elbow" inward at the 3rd switch-- which u cant do with ur other fingers. so i bet independent North & South directions for a 3rd switch within a finger triplet would be really difficulty except for the thumb. even if it were possible, i just imagine it'd be slower for the fingers vs the thumb for managing 3 sticks. so maybe the endgame CC device would be one with an extra switch for the pointer finger. even until that happens, if ever, we may still be able to make it work with the current CC2 and could tweak the layout a lil down the road if they ever add that extra index finger stick.

#

not to mention, having a 3rd bank could be clutch for making phrase chords

wide raft
#

Sounds like my Controller theory

Two thumbs do all the complex stuff
joystick + triggers + joystick

And I'll use the evolution algorithm to find the optimal chords

But with vastly fewer keys than a CharaChorder I'll be using multistrokes

MEN/-SHN -> mention
(Assuming "shn" is common enough to get its own chord)

IMG_20260112_000714.jpg IMG_20260112_000641.jpg
#

The reason there's an outer circle is for clockwise/anticlockwise movements

#

Which is something you said you find easy, perhaps could add to your theory?

#

So each joystick is 8 directions, + 8 clockwise + 8 anticlockwise
Hoping that's enough chords to write most words in not toooo many strokes? But obviously more stroke intensive than a CC or WSI theory

molten geyser
#

So do you go straight out for the inner circle? And for the outer circle, you go straight out initially, but then rotate a certain direction to indicate, which of the two you want? What if you were trying to go straight out but did not go perfectly straight out (by accident)– – can your software distinguish that from an intentional rotation?

#

seems like this controller theory has put so many restrictions on you that it forces u to be creative😎 . creativity often emerges when u have lots of restrictions put on u. cool project

#

maybe for certain chords in a CC2 steno theory, we do a rotation at the end to add on some additional information. it's not really 2 strokes-- it'd probs be a bit faster than having to go back in the middle and push the switch in a different direction. isn't doing something like that called apprepegiation? or is apprepegiation considered where u only press 1 additional key after a chord. other wise, if u press multiple other keys, u call it a compound chord i think.

not sure how that would work in terms of how the key presses are treated. a while ago i ask this question in #general. further below is the response i got, which i honestly forgot what exactly it meant:

Quick questions:

So you know how you have to release all the keys you’re holding down at once to initiate a proper chord? How exactly is “release” defined? Does that mean return back to neutral, or does that mean change directions in any direction e.g. with rolling? Like what would happen if I were to roll 1 of the keys I had held down? Would the key I originally held get registered as part of the chord and then the key I just rolled to get appregiated (assuming just 1 key. Further, would 2+ keys being simultaneously rolled after a chord trigger a compound cord?) Or would the OG key not get registered as part of the chord but the key that got rolled to would be appregiated (assuming u put it back to neutral afterwards)? Also, can u chain appregiations/compound chords? E.g. maybe I chord something, but then I roll the switch thru 2,3,4,5+ (you could even go in a circle) other directions before returning back to neutral? Would all of those trigger consecutive appregiations (or compound chords if u have multiple simultaneous long rolls)? E.g. rolling 1 key maybe appends an -ed to the word, 2 would append -er, 3 would append -est, 4 would append -ly. Btw Does rolling kinda slow down if u start going in circles or is it just as fast/satisfying as rolling only between 2-3 directions?

#

preferably we could squeeze everything needed for a 2 syllable word in 1 stroke, but if that proves impossible id probably settle on ["1.5" strokes that are easy to remember+have lower likelihood of contradiction] over [1 stroke rote-memorization brief with higher chance of contradictions] or over [2 strokes]. maybe that's smth we'd do for certain tough 2 syllable words or maybe even to squeeze a 3rd syllable in for potentially 1.5 strokes for a 3 syllable word

one thing i will note is that rotating 2 neighbor fingers in opposite directions can (but not always) be challenging

wide raft
# molten geyser maybe for certain chords in a CC2 steno theory, we do a rotation at the end to a...

I'm gonna use steno terms to describe arpeggiation cause not sure CharaChorder has them

chord:

a unit of meaning, TKPW → "g", S → "s" -PLT → "ment"

a stroke:

From the first key press, till all the keys are released
AR/PEPBLG/KWRAEUGS → "arpeggiation"
this is three strokes, as the hand was fully lifted 3 times

arpeggiation

Adding to a stroke one chord at a time
HERT|-Z → "hertz"
in this example the right pinky hit -T then -Z (these keys are diagonal, you can't physically hit them together)
this doesn't count as two strokes, because the whole time, all the other fingers remained on their keys

#

I will be using arpeggiation in my controller theory, cause I might not have space for something like a dedicated -nd chord, and I need a way to do suffixes like {^ed}
BAN|-D → "band"
BAN/-D → "banned"

#

So my software can distinguish strokes from arpeggiations, but it isn't doing this based on "time since last pressed" or anything, Plover actually can't see timings, instead it's based off if the chord was begun before the last one was released or not

wide raft
wide raft
molten geyser
#

@wide raft I've been doing the math. It seems like a 5 bank setup (3 consonant banks 2 vowel banks) would probably be impossible without some weird tricks. even if we somehow made it possible, the strokes for 2 syllable words would be really heavy probably (at least 1 finger per bank so 5+ bare minimum, more likely to be ~7 fingers if i had to guess. it'd be likely that we'd have some awkward strokes all the time). My next idea was only 4 banks-- 3 consonant banks and 1 vowel bank, and we'd just drop the unstressed vowel from the outline or whatever u call it. According to chatgpt (who did some coding analysis on a corpus/corpora for 25minutes), under Magnum steno, ~10% of 2-syllable words can be made using the steno order trick. But with a 4 bank setup, that figure becomes more like 90% (92% of 2 syllable words have only 1 unstressed syllable (i guess it's possible to have 2 unstressed syllables). of those, 80% have no conflicts. In practice, most conflicts are for niche words, so that 80% is more like 98.4% in practice. The other 1.6% would need a modifier for disambiguation kinda like "**". So 98.4%*x92%= 90%. lemme know if those numbers sound right).

So yeah. Now i've just been figuring out which thumb sticks to use for which bank. Chatgpt says that of the 128 (2^7) possible combos of onsets possible 61-64 combinations are actually used. For codas the number is more like 88 (88 could be wrong needa double check) i think it said of the 1000 total possible (2^10 or something like that), excluding the astrieks key.

how do onsets and codas work in 2 syllable words? do u think we'd need 1 onset bank and 2 coda banks or 2 onset banks and 1 coda? does it depends on the word...?

marble delta
#

tttttatatatttttttttttattttttttttaattatatatattttttttt

shut irisBOT
#

GG @marble delta, you just advanced to level 4!