>>8116 I probably should have posted this link earlier. Here's the relevant Google doc for the Pony Preservation Project from /mlp/. https://docs.google.com/document/d/1xe1Clvdg6EFFDtIkkFwT-NPLRDPvkV4G675SUKjxVRU/edit# This document is full of useful information, so feel free to peruse it. I used Deltavox RS to generate the audio in the .webm, and it's fairly straightforward to use. You don't need any coding knowledge at all. There's a user guide in the Google doc as well as a download link. You can try experimenting with some different input lines and parameters. I'm not really a brony either, but it's my hope that this proof of concept will encourage anons to work on similar projects with other characters.

>>8118 Thanks! BTW, any chance you could post the file(s) somewhere like catbox? If you wouldn't mind that would be a big help to us?

>>8118 Very cool, anon! Thank you. Gonna get hold of some MLP episodes and follow the tutorials. This project will be useful for anyone wanting to learn how to make their own synthetic voices instead of just relying on commercially available ones! A fundamentally important robowaifu building block!

>>8123 No need, there's already a mega link for Deltavox RS in the Google doc I posted: https://mega.nz/file/CMBkzTpb#LDjrwHbK0YiKTz0YllofVuWg-De9wrmzXVwIn0EBiII It has everything you need. Start it up load the model, pick a character and an emotion, and you're good to go. There's more information in the guide: https://docs.google.com/document/d/1z9V4cDvatcA0gYcDacL5Bg-9nwdyV1vD5nsByL_a1wk/edit >>8128 Just by the way, you won't need to worry about collecting any audio or training any models; all that has already been done. If you're planning on listening to the characters so that you can familiarize yourself with their natural speaking voice, then by all means go ahead.

>>8129 Oh, now I get it! I was just looking at the old contributions spreadsheet for each episode and it looked like everything had already been labelled and transcribed up to season 9. Amazing amount of work has gone into the creation of these voices. This should be in the media or something to get it more recognition!

>>8113 I see what you mean about having to play about with the pronunciations a bit, but I often have to do that with Sophie for certain words and phrases so not unexpected. I was blown away by how these voices can show different emotions convincingly.

>>8135 top kek.

>>8135 Guys I know we're having fun and all, but this conversation is certainly worth focusing in the dedicated thread for the topic. >>199 >=== -add topical crosslink

>>8113 Thank you very much for the resource links below Anon, and welcome back.

>>8140 I am looking into writing lines for characters in the game...but on the SPNATI website it says "Unfortunately, the audio files needed for music and voice acting would take up even more of SPNatI's limited space than the sprites do. Since the game is silent..." I am confused if the devs will allow this? Even so, just learning about 15.ai has increased my motivation. The guys over at /mlp/ have created an incredible range of voices here!

>>8140 Would you prefer to keep this discussion in the speech synthesis thread, or would it be appropriate for its own thread? I only made the first post here because I felt that it's more of an /hgg/ topic than a /robowaifu/ one. >>8142 Given the sheer number of characters in SPNATI, it would be very impractical to voice each of them. However, the intention of this project is to specifically exclude all characters other than the 4 MLP characters. That will dramatically reduce the storage needed. The audio data itself should be manageable. I've estimated that, using .mp3, the total file size should be well below 300MB. The devs have licensed the project under the MIT license, so it will be perfectly acceptable for me to create a fork of the project and change it as much as needed.

>>8149 >Would you prefer to keep this discussion in the speech synthesis thread, or would it be appropriate for its own thread? For now, I'd say keep it in the speech synthesis thread. If it gains a lot of traction here then we could consider otherwise later. Also, I'd gently point out to you that we are a SFW engineering board so please keep that in mind, thanks.

I've been lurking for a while here and I have to say that I really like what's going on! There's a really dedicated community as well as some very talented people involved in a project of this scope. Found this place via /g/ about a year ago but recently came back to it from my bookmark folder about a month ago. Seems like some significant progress has been made despite Corona's best efforts and it's looking really promising. This might be an absolutely awful idea, but I was thinking of attempting to start a "Counter Embassy" over on 4/g/. ( /rwg/? Maybe merge with /csg/? ) Whether this is a good idea given the current state of halfchan right now is up for debate; Shills are prominent, plenty of election tourists migrating to the hobby boards, and a few boards are borderline unusable even on a good day. It could either bring an influx of new blood which might be good for the creative process or it could bring bad publicity and people looking to cause this project harm. Just some thoughts from an interested engineer who would love to see this project succeed.

>>9148 >Found this place via /g/ about a year ago Welcome 4/g/, thanks for the encouragement and for sharing something about yourself with us here. Interestingly I began /robowaifu/ after a few ongoing threads with other anons there on your home turf. These took place probably 6 years or so ago now. So yeah, nice to see you here with us. I think we're quite likely to see our first basic home-grown robowaifu systems coming from here by this year's end, maybe even by our birthday celebration. >It could either bring an influx of new blood which might be good for the creative process or it could bring bad publicity and people looking to cause this project harm. Honestly, we've already been prepping for onslaughts when they occur -- both good ones and bad ones -- and have a few backups in place if worse ever came to worst. So, go ahead and share the board as you see fit. I think we'll be good to go. >Just some thoughts from an interested engineer who would love to see this project succeed. Out of curiosity, what kind of engineering Anon?

>>9154 >by this year's end I'll be looking forward to seeing what progress is made. > I think we'll be good to go. Neat, I will put up a thread after I make a general template in a few days. >what kind of engineering Anon? I'm an embedded systems engineer, thanks for asking. Do you have any scientific background?

>>9196 >I'll be looking forward to helping advance what progress is made. FTFY, haha. ;) >Neat, I will put up a thread after I make a general template in a few days. I don't think many of us will visit, but if you'd like to, post screen caps here for us in our /meta thread >>8492 . I'm sure we'd like to see your interactions there. > Do you have any scientific background? Well, I guess this is the Embassy Thread after all, heh. I'm very interested in science in general, and I'm a software engineer (though self-taught). I've had an interest in creating an 'AI Friend' since as long as I can remember -- at least by first or second grade at school. That desire turned out to be a big part of equipping me to fill the role of founding this board. It seemed like it just had to happen, no question. I'm sure others might have served just as well, I just happened to be the one.

>>2823 I'm an anon from /mlp/ and I'm fucking around with a plushie. >pic related Is a mockup of a little project I'm doing.

>>10983 Welcome /mlp/. Interesting project, good luck with it. There have been a few ponies visiting our board from time to time, as can be seen ITT. Is that 4/mlp/ , BTW? There has been some interesting work by some of the pony anons regarding synthesizing the VA voices (>>1563, >>8113, >>8118), ostensibly to 'preserve' them. It's actually been a rather sophisticated effort IMO, and evidences some of the benefits of fan dedication.

>>10984 Thanks, and yes I'm from the 4chan board >>>/mlp/. I've done zero work on the voices, but I figured a physical platform like this could use the excellent voices and the beginnings of a chatbot made by other anons. I've posted on the actuator thread here for over 6 months now. My plan is to make a 12 inch plushie bot, then 36-50 inch plushie bot, then something similar to Spot the robot dog, and then a plane-pony (plony) for the lols and to increase my powerlevel. I'd recommend another anon to corner the likely lucrative furry or FNAF robot market, which I cannot stomach. It's really incredible what autism can create.

>>10983 >>11024 Maybe you should have a look at this project Anon (>>11122)

so here is my regurgitation shit pile thought . AI first module is for start up. on start up, creates a seperate drive/file, it creates a randomized labrynth matrix architecture within the computer. stage two, self learning activates. this is where the ai has restricted building abilities butit builds more ordered eldritch architecture . . you give it a folder full of stuff you want it to hot box echo chamber. or not. you could also just have a folder of all basic stuff and scatter in the environment (btw this will likely be a virtual machine sorta thing) ai then learns, spl.its, wages war, evolves. then we get to watch the progress and th end result. we could even have a turn off point measure to stop it getting too powerful

>>11124 More here >>1563 and following, plus >>8295 >>10470 >>10492

how much money would be needed for some of you lot to put together a quality AI system? im talk the best you could possibly do without getting spooked in the night. not waifu per se, more like ideological or maybe even just companionship. shifty#9034 is my discord btw (dont have much money so will probably just be friends for a while.)

>>11234 It's not about the money, but skills and effort. No one can guarantee you that something good would come out of it or how long it would take or what it would cost. Of course if I had some millions to spent I could hire some people and give it a try...

>>11234 Hello Anon, welcome. As the anon pointed out to you, there really are no guarantees about devising a quality AI system. There are example chatbot AIs like GPT-3 that took many millions and years to create. There are also stringent hardware requirements for that statistical approach as well. So, making wild presumptions on both the meaning of the word 'AI' and our progress as a race on this truly frontier area, I'd guess roughly US$ 20-50 Bln, and maybe 20-25 years from now? Have a look around the board while you're at it, there's lots of discussion on this topic in general here.

>>11241 even GPT-3 is just a fancy way to link database string entries am I wrong? is anything even "thinking" in such a scenario or is it like someone talking in their sleep? Has anyone yet tried to combine this chat "function" (which would be only part, or one "function", [imitating human chattiness] of a true AI consciousness) with persistent memory, ego emulation, visual processing, and the laundry list of what would actually be necessary for an autonomous being which could pass for self-aware? Replika AI tries this but it feels like emulation with a few simple "factoids" it stores about you. tl;dr anons dont want to fall in love with a database, anons want to talk to something that processing their information and perhaps even forms a model of them in its "mind"

>>11241 >I'd guess roughly US$ 20-50 Bln, and maybe 20-25 years from now? Lol, that's not very optimistic. What are we even doing here, then?!? Within the next 5-7 years we are going to have something decent, and it will show before that. Just not with unnecessary requirements like "It has to be trained from the scratch, no other programs" ... "It needs to be a black box with a will of it's own" ... >>11242 I don't know how much is has been combined with something else, they might not make that public. OpenCog might have.

>>11235 i guess its more fair to ask for something equal to what i can pay

>>11242 >>11234 haha I love the ending on that sentence. I think high quality is relative if you checkout my thread, the 4 model is what people prefer >>11242 I think maybe that's what you want but most people lean towards the simpler 2010's chat bot technology. I think as conversational ability degrades we will converge to AGI because people will be simpler than AI's in the current form.

>>11242 I certainly can't claim any expertise on the topic Anon, but as I understand it, the transformer simply uses a statistical probability model constructed during it's training phase to make a weighted prediction of suitable following text regarding input strings. This is why it works from an initial 'seed' &tc. And no, it most definitely isn't 'thinking' in any human sense. I'm personally skeptical that will ever happen, quite frankly. >anons dont want to fall in love with a database... Heh, very well-put Anon. I think we'll get there eventually with suitably satisfying simulacrums, just be patient. >>11243 >Lol, that's not very optimistic Fair enough Anon. :^) My estimate is on what we would roughly call an AGI today, and on second thoughts, I suspect my cost estimates are at least an order of magnitude low. >Within the next 5-7 years we are going to have something decent, I think you are correct Anon. Certainly something well-enough suited for a 'starter robowaifu' if you will. >>11299 >I think as conversational ability degrades we will converge to AGI because people will be simpler than AI's in the current form. LOL. I think just the opposite Anon, though I'm well aware of both the dynamics you're implying. Humans are profoundly capable and multifaceted. By far the most spectacular of all the natural creatures!

>>11321 >what we would roughly call an AGI today, >>Within the next 5-7 years we are going to have something decent, >Certainly something well-enough suited for a 'starter robowaifu' if you will. I'm confident that within that timeframe it would be possible to get something close to a human in many if not most cases, within a household. There might be cognitive gaps, where the system fails and can't kearn the right pattern, but I don't see why fixing and polishing this would cost billions or trillions or being impossible. Having that said, best way to find out is not to waste time and get to work. Thinking about this, at least got me back to learning and working on it. Five years will pass way too fast.

>>11361 >Having that said, best way to find out is not to waste time and get to work. THIS!111 :-DDDD

Have any of the anons here posted on 8/vg/ back on 8ch? We're trying to make a board revival with a spiritual successor. https://anon.cafe/valis/catalog.html If you haven't, 8/vg/ and now /valis/ is a comfy vidya board for actual game discussions. Come check it out.

>>16054 Hello /valis/, welcome! Thanks for the invite and for letting us know a little about yourself. BTW, I personally consider your board to be the single best vidya board on the Internet today. I applaud your regrouping on the Cafe, and I wish you blessings with the future of your board. Thanks for stopping by, please look around while you're here! Cheers. :^)

>>16055 Thanks fren.

>>11235 >It's not about the money, but skills and effort and that's about 100k $ per year per engineer- assuming they are driveny by that goal, because it'd be a low wage for a competent ML/DL/AI engineer.

This seems like the best thread to ask, so: what is your opinion on techpriest esque waifus as opposed to a full robot? As well as becoming one yourself? It seems to be way easier to me, and you filter out roasties by the fact that only a woman of character would go through something like that (or innocent vat grown humans whatever). If you became a techpriest yourself would the likely lack of sexual ability dissuade you, or would wholesome lovey Dovey roamnce and faith shit be enough?

>>16643 Also personally I think AI is evil unless you viably and carefully transition humanity into it.

>>16643 I'm pretty big on transhumanism myself, but I don't see a perfect woman coming out of the tech priest program. >>16644 I am partial to the idea of using xenocyborgs with biocomputers for brains personally. It should be easier to get something capable of processing multiple parallel processes at once to think like us than traditional AI.

>>16643 https://www.youtube.com/watch?v=mVLEJ9awLos makes me think of 4:35

We at mlpol.net/cyb would like to extend our hooves in friendship to the cause of building robowaifus. Mlpol.net is a /pol/ offshoot that was formed after an April fool's joke in 2017 combined /pol/ and /mlp/ on 4chan. We also have a cyberpunk board /cyb/ and those of us on it have a vested interest in creating robowaifus. Just think about how many engineering problems would be solved by simply making quadruped robowaifus!

>>17623 Come to think of it the tails could help with balance as well.

>>17623 Hello /cyb/, welcome! I'm personally aware of mlpol, but as they didn't allow TOR posters in the past I didn't participate much. I'm relieved to see that's changed today. Regardless, you have some good points and the concept of quadrupedmore-than-two-legs robowaifus (or just wheeled ones) has been bandied about here before. It's a good place to start, at the least. BTW, there are other active pony communities that occassionally engage with us here on /robowaifu/ as well. Look around you might find some other new frens too! :^)

>>17623 I posted there in the past, I think (the AT&T comes to mind). Not really into MLP myself, did it just to keep the connection going. But they seem not to try building something, just waiting. Anyways, welcome (again?).

>>17625 What other pony related communities come here? I could be shilling on those places too!

>>17698 heh you might try looking around nearby anon, who knows? >ctrl+f 'pony'

>>17623 Hey /mlpol/, anon from /mlp/ here. I'll check out /cyb/. I can't believe I never came across it until now. On /mlp/ we have /ppp/, which contains many resources. Most of the ai bots off board are anthro furry trash, and each month it seems like there's more and more pony bots of all kinds to play with. The future is bright for the autistic horsefucker.

Corechan is mostly about technology. There's an off-topic board too. Come and check it out. https://corechan.net/index.html

Alright, I'll merge this thread into our Embassy thread, OP. I'd urge you to keep spam and other chaff away from your site--it takes a little diligence. :^) Cheers.