Oh yeah this def the right curve finally.
Dong Juan
noise is the best part
that’s where da seductive demons hide
1 Like
They tell me to gamble optionz to make back my losses!
I had an interesting idea. 78rpms are an encode decode system and there are very real physical limits to what a stylus can do during encode. Thinking of modeling what these are and creating an exclusion criteria to restrict all outside events. If it’s physically impossible for a stylus to have encoded it, probably surface shits and damage.
2 Likes
that’s a really interesting idea. i’d love to see what you can do with it!
It failed sadly. There were two main issues, 1. I didn’t have finite enough info on the physical characeristics of encode so had to use general estimations, which I got modeled ok on a theoretical groove wall I mapped constraints to, BUT 2. the playback side is rattling around pretty crazy and smears/mixes critical info you’d use as a differentiator between an encoded or non encoded event due to it’s own physics.
By the time you are done you are stacking several imprecisions and randomly changing read info layers and nothing really flags as high confidence. I bef if you could impulse response both encode and decode it would kind of work tho. That way you could generalize things and remove tails a bit to get a picture of what’s going on at both layers.
Randomly this could be interesting for optical playback analysis but I dun want to contact those scientists cuz lazy.
i am building this, gonna be most involved click removal program on earth. YESSSS
Gonna take a couple days or I just redbull through the night until I fill those pseudo holes OH YEAH!
I legit read that as something else removal
1 Like
diz complex as shit, fuck. A lot of these ideas dun go cleanly together like I imagine when doing pictures in squares and not codez
1 Like
Please find da next 10x stock for moi n trupops
Probably people who sell fallout shelters and iodine. They are thinking about dropping nukes on Iran because they won the latest Israel war we are doing.
1 Like
Hyper accurate click removal starting to work. Is spooky. Details no longer smeared and surface noise sounds sort of like raw just without clickz.
Will see how far I can push this shit.
ur not bitter lesson pilled ?
1 Like
I like deterministic shits for clickz cuz I can tune them on synthetic noise / signal pairs at both identification and repair layers without ambiguity. That way I can isolate events better and protect more ungaped signalz bitz
Got a long way to go tho. Happy with repair methods but identification may be too strict. RX finds way more clicks generally even if repair not as advanced.
In perfect world this go first and reduce need for mixing a lot of UNET Moliner processed stuff in which separates random noise better but soften everything too much generally.
1 Like
Ahhh I figure it out. Periodicity shits was vetoing half my identified and otherwise listed as non musical event clicks. Fixing now.
Ok this have 23 thousand clicks removed perfecto way but there some literal chunks missing so used the deconstruct thingy to take out mega divits this miss.
NO AInus or broadband stuffs.
Is ok but poss need better transfer and to mix wet with an AI cleaned version.
1 Like
Diz lovely. Rach wuz good at da 88.
1 Like
hell yeah. Rach most def on of da most original 88 playas evah
2 Likes
Randomly…i kant say i heard you play any Rach at all.
I kinda wanna c tromofo 33/1 you nigger
Perhaps a cummentaire 