Welcome to the Pony Voice Preservation Project!youtu.be/730zGRwbQuEThe Pony Preservation Project is a collaborative effort by /mlp/ to build and curate pony datasets for as many applications in AI as possible.Technology has progressed such that a trained neural network can generate convincing voice clips, drawings and text for any person or character using existing audio recordings, artwork and fanfics as a reference. As you can surely imagine, AI pony voices, drawings and text have endless applications for pony content creation.AI is incredibly versatile, basically anything that can be boiled down to a simple dataset can be used for training to create more of it. AI-generated images, fanfics, wAIfu chatbots and even animation are possible, and are being worked on here.Any anon is free to join, and there are many active tasks that would suit any level of technical expertise. If you’re interested in helping out, take a look at the quick start guide linked below and ask in the thread for any further detail you need.EQG and G5 are not welcome.>Quick start guide:docs.google.com/document/d/1PDkSrKKiHzzpUTKzBldZeKngvjeBUjyTtGCOv2GWwa0/editIntroduction to the PPP, links to text-to-speech tools, and how (You) can help with active tasks.>The main Doc:docs.google.com/document/d/1y1pfS0LCrwbbvxdn3ZksH25BKaf0LaO13uYppxIQnac/editAn in-depth repository of tutorials, resources and archives.>Online speech generationhaysay.aialpha.15.dev>Active tasks:Research into animation AIResearch into pony image generation>Latest developments:pastebin.com/4p00iUZM>The PoneAI drive, an archive for AI pony voice content:drive.google.com/drive/folders/1E21zJQWC5XVQWy2mt42bUiJ_XbqTJXCp>Clipper’s Master Files, the central location for MLP voice data:mega.nz/folder/jkwimSTa#_xk0VnR30C8Ljsy4RCGSigmega.nz/folder/gVYUEZrI#6dQHH3P2cFYWm3UkQveHxQdrive.google.com/drive/folders/1MuM9Nb_LwnVxInIPFNvzD_hv3zOZhpwx>Cool, where is the discord/forum/whatever unifying place for this project?You're looking at it.Last Thread: >>42429020
New vibe voice out, anyone try it?
which one
>>42973521>>42972932
oh shid, are we back?>https://huggingface.co/Edge-Quant/YuE-s2-1B-general-Q4_K_M-GGUFDoes anybody know how were to get a ui to run this thing? It's two months old quantize model of the YuE music making model (original one was 5GB, this one is little bit over 1GB)?
new tts with voice cloninghttps://huggingface.co/Qwen/Qwen3-TTS-12Hz-0.6B-Basehttps://huggingface.co/Qwen/Qwen3-TTS-12Hz-1.7B-Base
>https://github.com/kyutai-labs/pocket-tts>https://kyutai.org/blog/2026-01-13-pocket-tts>kyutai finally released their Pocket TTS last week>runs on CPU and has voice cloning>supposedly is faster than real-time audio generatingFinally these French fucks had finished pussyfooting around. I haven't test it but I would imagine it would be pretty useful to as narrator/character voice in the text gens
>>42973675>100M-parameteri really doubt it can handle pony voices. Even the timbres, let alone speech patterns. 1.6B vibevoice struggled and 1.6B isn't even the whole network
What exactly caused the death of PPP? It was pretty awesome when generative AI was a novelty and we were way ahead of the curve
>>42973700>when generative AI was a noveltyIt stopped being a novelty and voice models stagnated so much with no news that most anons that aren't hardcore into it just stopped bothering.I know many blamed it on the death of 15, but he came back and that just gave a minor boost for a short while, because his voices are at the same quality as the rest, aka not that great and not evolving much since years.
>>42973722Makes sense that it's because development has kinda slowed a little. This shit takes a lot of work; there are some promising local video models coming up at the moment but as far as pony development goes we'll have to see if anything picks up for it.It's exciting nonetheless, truth be told.
>>42973700I blame the dumbassess that code + make the models requiring ALL the ram and ALL the newest modules without testing whenever their new code works at all>-t guy who is trying to install torch 2.5.1+cu118 that should work on gpu but doesn't just because fuck me I guessMy personal hopes are that we will get another Deepseek situation were Chinese devs will figure out how to run a good quality model on potato pc.
>>42973693hmm, I wanted to try it, as I still remember their website working pretty well few months back, but their main website isnt loading the model, and their hugging face space web thing is giving me trash random noise sounds as output .
>>42973521>>42973523Vibevoice asr
>>42973920>generating structured transcriptions containing Who (Speaker), When (Timestamps), and What (Content)That's pretty neat, specially it supposedly works with hour long audios. Does anybody know what a text-to-text version of that would be called (like model loads a text novel, and separates the lines between what is being presented by the narrator and what's spoken by the characters) ?
>>42974102Seems like a given there are plenty of examples of screenplay versions of texts which segment like that
>>42974109my googlefu seems pretty weak, since all I see is people shilling chatgpt/grok/cloude with extra instructions and not a properly trained novel-to-screenplay model. I could take the easy way like that however I don't trust these services to not hallucinate some bullshit in between the lines (and also local LLMs that are somewhat as good as they are require two 5090 to even start up the bloody model).
>>42974152I wonder what use a dedicated screenplay conversion ai could serve PPP, we already have the transcripts from the wiki, and more significantly the leaked official scripts including all their revisionsI suppose it could be helpful in translating fanfics into animation-ready format that remains consistent with the show
>>42974173>translating fanficsthat's exactly what I was thinking, while tts models that are combination of local, light weight and great sounding doesn't exist just yet, I thought it would be neat to have something to segregate the text lines into their own character lines and feed these lines to the tts model for a simple audiobook production.
>>42973675Just testing this one, their example voices are OK, better than MS Sam but obviously not very emotive. I've tried using their cloning setting but it looks like its connected to their huggingface account to make sure it's not use for "the wrong thinking" purposes.I see the github has a training text file, with links and tutorial how to run it, but with the lackluster ui that came with (and looks like the training script only works on Linux or the win10 WSL), which is a shame because I would love to see a tts model with cloning that is as simple as "press button, get voice" as this would be a dream system set up to work with.
>>42973700Everyone involved that actually did things stopped doing things.
>>42973700AI voice tech is only one infinity stone out of the six you need to make AI content.The race now is to see who can make the first twenty minute MLP episode, not who can make the best AI voice.This is a shoddy example and in 3D, but see I made this with a custom voice that I crafted using Voicelab in Voicemod: https://tantabus.ai/images/67885I'm excited for them to release their prompt to voice feature that they are working on which will be huge for making unique voices for OCs.
>https://github.com/index-tts/index-tts>https://www.youtube.com/watch?v=u2fGWqaOizk>https://huggingface.co/spaces/IndexTeam/IndexTTS-2-DemoThe Chinese are at it again, if not for the fact that it requires at minimum a 10 vram to work with it, I would be almost willing to try to shill it to other Anons (also on semi related note , holy shit, HF are really grinching on the demo space, only managed to make two clips before it cut me off).Btw, thank you HydrusBeta for keep the haysay alive, I may not use it as often as I did in the past but it sure does help to have it available 23/7 when its really needed.>>42974561Life happens, and especially with everything costing at minimum double, to get a bare bones pc to even just try the local models is disheartening (and with the upcoming "hardware as the cloud service" bs the nvidia and other companies are talking about, future looks grim).>>42974705>https://files.catbox.moe/8tiu8f.webmJust before new year A dedicated Anon in the ai image thread made this one minute ai dubbed music video. It was stitched together from three dozen edited short clips (and who knows how many failed clips were involved in pre production), so it is possible to brute force a episode if somebody was willing to lock themselves in a super-neet basement and do nothing but generate short 5~10s clips for an entire year for the raw footage that would need another year to remaster into actual episode.With how all the ai video models progressed from just animating 6 frames to a neat 10s, I have high hopes the tech will develop enough for making proper pony episode by the time PPP has its decennial anniversary.
>>42974762Oh yeah I made that too. It took me a week to get 28 decent scenes. Like 3-4 days to get the clips and 3-4 days to process upscaling with a rtx 4090.That's the 30 fps version from Tantabus because I was limited by filesize. The 60 fps version is on civit: https://civitai.com/images/116777810I used so-vits-svc for the voice because it was the best Diamond Tiara voice I could find.It really hurts to see my smut get more likes than what this got, but what can you do.
>>42974786THE FOUNTAIN MARE LIVES
>>42974804There originally was a fountain mare but I had to remove it because wan 2.2 is insistent that statues are people too. In the future I'll have to use a corpo model like grok to animate scenes like that.I'm tempted to make a directors cut with only the worst clips.
>https://files.catbox.moe/otlaa2.mp3
>>42973662>1.7B-Base(d)(3) Twilight: https://vocaroo.com/14WL8R4v7Vod(2) Rainbow: https://vocaroo.com/1j10R0C5gWhS(2) Pinkie: https://vocaroo.com/12OXKO0D4rRl (1) Rarity: https://vocaroo.com/16e7dSGEv9Jk(3) Applejack: https://vocaroo.com/1dyLZxLnNww3(2) Fluttershy: https://vocaroo.com/1oThBeZkNePLNumber of rerolls between parentheses. It sucks that providing a reference text works better than without it, so I have to use whisper to automatically caption the reference audio.Overall, to me, this feels like a sidegrade to Vibevoice 1.5B when it comes to voice cloning: Qwen cloned Pinkie's voice a lot better, but it performed worse with whispers (ASMR (Not showint that, though...)). It also felt like some output voices were a bit deeper than they should be. I still think Vibevoice has more emotion, but that might change if Qwen gets to add "voice control" to their voice-cloning model.
>>42975272Oh yeah, my results are pretty quiet for some reason, so turn your volume up if you don't hear anything.
>>42975272gptsovits not going anywhere
>>42975272>>42973662uhhh, do this tts have any webui or do you just use the raw console input to operate it?
Yo, would Anons here be interested in doing a little challenge creation, of making a ai generated songs based on a weekly/biweekly theme?
>>42975509According to their github, it has a Web UI: https://github.com/QwenLM/Qwen3-TTS?tab=readme-ov-file#launch-local-web-ui-demoBut, I just used raw console inputs to generate those examples
>>42975272>>42975995Interesting, like the Anon said I would probably stick with the gpt-sovit just because I already have that installed however it could be interesting to test the voice cloning for characters with just one/two lines.
>>42974246I'm working on this, it's still raw but shouldn't take long to finish>https://files.catbox.moe/za2c79.mp4The video is not sped up. LM used is qwen3-4b via llamacpp, but you could plug in anything. Don't mind that the labels don't match the voices, I've hard-coded rarity as narrator and trixie as everyone else while testing but there are finetuned models for most characters out there (therealvul/GPT-SoVITS-v2).
>>42976577IMO thats more or less what I was looking, even if it was limited to narrator + one voice over character it would still be pretty great for making diy audiobooks (as an autist that used the basic bitch microsoft tts to listen to a 10k worded fics, only thing to use anything else is the fact that everything is capped at 10s clips and constantly glitches out on random sentences).I still have hopes that the ideal version of a screenplay format convector script to be made in few years that also is able to add appropriate background sound effects and musical incidental/leitmotif to play as character talk.
>>42976780Any half-decent lm can do annotations like this but i'm afraid it would take much longer to process than just determining who said what. And the model needs to be bigger.
Cool, PPP is back. Here are some random bits of pony ai covers found on net from few months back. Whoever made them was clearly not doing best job when it comes to reverb and pitch editing, but still, its kind of neat.>https://files.catbox.moe/sd7c0h.mp3Applejack - Fowers - Miley Cyrus>https://files.catbox.moe/o3g4e7.mp3Applejack - When She Loved Me (Sings AI)>https://files.catbox.moe/gv9v2m.mp3Applejack - You Belong With Me - Taylor Swift>https://files.catbox.moe/t03xcv.mp3 Fluttershy - A Thousands Years>https://files.catbox.moe/zced52.mp3Fluttershy - Clarity - Zedd ft Foxes>https://files.catbox.moe/ogtfkm.mp3Fluttershy - Everytime We Touch (ai cover) BOB>https://files.catbox.moe/9wkk1n.mp3Fluttershy - I Need Your Love Calvin Harris ft Ellie Goulding>https://files.catbox.moe/m4brvv.mp3Fluttershy - Orange show speedway by Lizzy McAlpine>https://files.catbox.moe/ct0476.mp3Fluttershy - Roar - JaqeyMaq>https://files.catbox.moe/ujwr02.mp3Fluttershy - Yellow - Coldplay>https://files.catbox.moe/f9a46y.mp3Glimmer - Bawa Dia Kembali - Mahalini>https://files.catbox.moe/1xjn4m.mp3Luna - I will Always Love You (Ai Cover) Norsewarrior>https://files.catbox.moe/ltmlzy.mp3 Pinkie Pie - Baby One More Time (Sings AI) BOB>https://files.catbox.moe/xu8vt2.mp3Pinkie Pie - Oops I Did It Again (ai cover)>https://files.catbox.moe/0gx18i.mp3Pinkie Pie - Sweet But Psycho - Ava Max>https://files.catbox.moe/4glpe6.mp3Rarity - Clarity By Sings (AI)>https://files.catbox.moe/yr14y7.mp3Rarity - Sings Solena Gomez I love You Like A Love Song>https://files.catbox.moe/u0s3qe.mp3 Super Shy - FS PP RD>https://files.catbox.moe/qi9msp.mp3Twilight Sparkle - Fireworks Katy Perry>https://files.catbox.moe/bvp47t.mp3Twilight Sparkle - Isyana Sarasvati - Tetap Dalam Jiwa (AI)>https://files.catbox.moe/ow9sew.mp3Twilight Sparkle - Speak Now Taylors version>https://files.catbox.moe/n2ejhz.mp3 Twilight Sparkle - Taylor Swift - Daylight (AI Cover)>https://files.catbox.moe/bw69vv.mp3 Twilight Sparkle - The One That Got Away>https://files.catbox.moe/djzm3f.mp3Twilight Sparkle - Cool Kids Echosmith>https://files.catbox.moe/0vdzy6.mp3Twilight Sparkle - Cupid>https://files.catbox.moe/wgkj8h.mp3Twilight Sparkle - It Ain't Me - Selena Gomez kygo ai>https://files.catbox.moe/fzef8e.mp3Twilight Sparkle - Sayap-Sayap CINTA - TS AI>https://files.catbox.moe/4o8ab5.mp3Twilight Sparkle - Teenage Dream>https://files.catbox.moe/uo0e8i.mp3Zecora - Dig A Little Deeper
>>42977481>>https://files.catbox.moe/qi9msp.mp3>Twilight Sparkle - Fireworks Katy Perryit's so fucking peak brosimagine 2012 fags seeing this, puts a damn smile on my face...
>>42977481>>https://files.catbox.moe/ogtfkm.mp3 (unembed)
>>42977481>https://files.catbox.moe/sd7c0h.mp3>https://files.catbox.moe/o3g4e7.mp3>https://files.catbox.moe/ogtfkm.mp3>https://files.catbox.moe/ltmlzy.mp3>https://pomf2.lain.la/f/om249yy.mp3>https://files.catbox.moe/qi9msp.mp3>https://pomf2.lain.la/f/p7l4s8jj.mp3
>>42975760I guess this kind of thing is massively lacking in organisation these days.
>>42978009HOLY FUUUUUUUCK
>https://files.catbox.moe/f6rfv6.mp3>AJ explains to Applebloom how to use the new oven
I'm working on a song cycle as an assignment for composition and thought it would be funny if it was secretly pony. Here's the first movement sang by Celestia (Luna didn't sound good).https://files.catbox.moe/nw03w6.mp4
>>42979412Kino art songI've not heard one in the fandom
>>42979412I don't know if I quite agree with the notation for the speaking section in mm 45It could be more clear In fact I would move the fermata to the next measure and have it on "mortifying!" I don't think this would cause confusion for the pianist because they are already holding the chord, and for the singer you could give a marking that further indicates its a spoken line/the tone of voice Bravo anon
>>42979412dunno, for something that's made from scratch it's pretty good, however im not really feeling the flow of melody (I dont have enough technical know-how to explain it other than it parts of it just don't vibe with my ear).The pitch of voice also not always comes through correctly but thats just something all male to pony voice conversion models have problems with.
>>42977593>imagine 2012 fags seeing thisThey'd go nuts over that.
>>42977481This reminded me of this video that I still have on my computer kek: https://files.catbox.moe/pgw3mx.mp4
>>42980821And this one, have some mareio: https://files.catbox.moe/5hrzo9.mp4
>>42979722>>42979761>>42980100Thanks for the kind words, Anons. Also, nice dubs you two.>42979722>I've not heard one in the fandomYeah, not many people compose contemporary classical music in the fandom. Especially not this kind. Closest I've seen is LYRAnonymous' Rhapsody and Serenade.>https://youtu.be/6Qmr56JGhyY>>42979761>It could be more clear Yes, I know what you mean. I was thinking about how to notate this thing for some time, and I think you might be right that it would make more sense to be on "mortifying!" instead. I will talk to some singers and my professor again regarding this.>a marking that further indicates its a spoken lineThere will be performance notes at the start of the score that differentiate between spoken word with defined pitch with rhythm, spoken word with undefined pitch with rhythm, and free spoken word. MM 45 and 46 fall under the third category (Text below the staff with just rests). Since this will be a 4 movement song cycle and only the first movement exists in its entirety, I didn't bother making it yet. There might be other things I'd need to specify after it's all done. And I honestly didn't think there'd be anyone here who would even think about this, so I found it redundant to make it just for this video for this thread. You surprised me, Anon.>>42980100>im not really feeling the flow of melodyYeah, Messiaen-esque and contemporary melodies/harmonies might sound strange to the unfamiliar. It's just a quirk of the style.>The pitch of voice also not always comes through correctlyI did pitch adjustments to most things, but it's true that I didn't put a massive amount of effort into it, plus I had little time. The song will be sung by a real singer from my conservatory (hopefully), so this mostly exists just for the general idea of what it should sound like. I am also a pianist, not a singer, so that affects the quality too.
>>42980821that link is broken anon, or at least catbox isnt loading up the video, could you reupload it plz?
>mared
>>42981162Oh, sorry. I tried again with catbox, but it kept saying it was corrupt, and pone.rs got stuck at 12%. Pomf saving the day:https://pomf2.lain.la/f/ks6fr95y.mp4https://pomf2.lain.la/f/e6vk61cm.mp4https://pomf2.lain.la/f/wefpvs2.mp4https://pomf2.lain.la/f/squtqz0t.mp4
>>42972267Where's the rest of the OP?
Scootaloo Sings Down Under Australia Day 2026 ai cover Special Enjoy Everyponyhttps://files.catbox.moe/jxk9q4.mp3
>>42984802kino
>>42984802Neato
>https://files.catbox.moe/fcoy6e.mp3Fluttershy lagtrain
>>42985945Nice
>>42982172What's missing?
>>42986732I think me meant the send OP post:FAQs:If your question isn’t listed here, take a look in the quick start guide and main doc to see if it’s already answered there. Use the tabs on the left for easy navigation.Quick: docs.google.com/document/d/1PDkSrKKiHzzpUTKzBldZeKngvjeBUjyTtGCOv2GWwa0/editMain: docs.google.com/document/d/1y1pfS0LCrwbbvxdn3ZksH25BKaf0LaO13uYppxIQnac/edit>Where can I find the AI text-to-speech tools and how do I use them?A list of TTS tools: docs.google.com/document/d/1y1pfS0LCrwbbvxdn3ZksH25BKaf0LaO13uYppxIQnac/edit#heading=h.yuhl8zjiwmwqHow to get the best out of them: docs.google.com/document/d/1y1pfS0LCrwbbvxdn3ZksH25BKaf0LaO13uYppxIQnac/edit#heading=h.mnnpknmj1hcy>Where can I find content made with the voice AI?In the PoneAI drive: drive.google.com/drive/folders/1E21zJQWC5XVQWy2mt42bUiJ_XbqTJXCpAnd the PPP Mega Compilation: docs.google.com/spreadsheets/d/1T2TE3OBs681Vphfas7Jgi5rvugdH6wnXVtUVYiZyJF8/edit>I want to know more about the PPP, but I can’t be arsed to read the doc.See the live PPP panel shows presented on /mlp/con for a more condensed overview.2020 pony.tube/w/5fUkuT3245pL8ZoWXUnXJ42021 pony.tube/w/a5yfTV4Ynq7tRveZH7AA8f2022 pony.tube/w/mV3xgbdtrXqjoPAwEXZCw52023 pony.tube/w/fVZShksjBbu6uT51DtvWWz>How can I help with the PPP?Build datasets, train AIs, and use the AI to make more pony content. Take a look at the quick start guide for current active tasks, or start your own in the thread if you have an idea. There’s always more data to collect and more AIs to train.>Did you know that such and such voiced this other thing that could be used for voice data?It is best to keep to official audio only unless there is very little of it available. If you know of a good source of audio for characters with few (or just fewer) lines, please post it in the thread. 5.1 is generally required unless you have a source already clean of background noise. Preferably post a sample or link. The easier you make it, the more likely it will be done.>What about fan-imitations of official voices?No.>Will you guys be doing a [insert language here] version of the AI?Probably not, but you're welcome to. You can however get most of the way there by using phonetic transcriptions of other languages as input for the AI.>What about [insert OC here]'s voice?It is often quite difficult to find good quality audio data for OCs. If you happen to know any, post them in the thread and we’ll take a look.>I have an idea!Great. Post it in the thread and we'll discuss it.>Do you have a Code of Conduct?Of course: 15.ai/code>Is this project open source? Who is in charge of this?pony.tube/w/mqJyvdgrpbWgZduz2cs1CmPPP Redubs:pony.tube/w/p/aR2dpAFn5KhnqPYiRxFQ97Stream Premieres:pony.tube/w/6cKnjJEZSCi3gsvrbATXnCpony.tube/w/oNeBFMPiQKh93ePqTz1ns8
>>42972267Neat
>https://pomf2.lain.la/f/odkytnn.mp3Luna - Don't Blame the Moon - IO PertyIm not exactly 100% happy with this one, specially when the duet parts are as neither rvc or sovits know how to handle these bits. I may try to salvage it , if I happen to have a free weekend.
>>42987023Yeah that's important.
Pony Friendship Hunters - Free (Sung by Twilight Sparkle & Rainbow Dash)>https://pomf2.lain.la/f/dv60hne7.mp3
>>42989243very nice Anon
>>42989243Is this based on any other song or 'original'?
>>42990737if I told you you would sack me
>>42977481damn, thats one heck of a nostalgia ride
>>42991006Go be a retard somewhere else.
>>42976577Finished the tts app. >>42992572
>>42992585oh hell yeah, new tts. What model is this based on Anon?>9GB 40m downloadwelp, looks like its going to be a big file
>>42992629good old gpt-sovits>big filebecause it contains all models therealvul's repo.
>>42992643>gpt-sovitsoh, cool than, I know I can run it without much of a problem than. How easy/hard would you say would be editing thecharacter list and adding some other custom character (since there is few models that were trained by other folks )?
>>42992675wait, is it the old gpt-sovits v2 or some other version?
>>42992681>>42992675v2.>adding some other custom characterYou'd have to:>create a folder in models for the new character and organize it like the rest.>add a folder to ref_audios_inference_server with the reference audio>add an entry to the dict in inference_server.py
>>42992585 >>42992715>x is shitting itself in half a second from openingwelp, time to run that python exe wtih VxKex on>Loading librieries>avutil-58.dll is missing>avutil-57.dll is missing>avutil-56.dll is missing>NFOutWorkItem2 could not be located in dynamic library kxdx>get pic related console error before auto shutdownfug, I was looking forward to testing that
>>42992832Interesting, do you have ffmpeg? Try installing it.
>>42992832Could you add these files to the dir wtih bat and try again? https://drive.google.com/drive/u/2/folders/16zjFqobYW7aQrU0w8QFAIYRPFKoRg6g5It's my ffmpeg, if it runs i'll include it in the main app
>>42992847I do have it, otherwise the audacity wouldn't work for me, here is pic related>>42992874>avutil-58.dll is missing>avutil-57.dll is missing>avutil-56.dll is missing>NFOutWorkItem2 could not be located in dynamic library kxdxand console message error:Loading libraries, may take a while...Initialized with max_loaded_models = 10Exception in thread Thread-2:Traceback (most recent call last): File "threading.py", line 980, in _bootstrap_inner
>>42992884Try installing https://aka.ms/vs/17/release/vc_redist.x64.exeAlso try replacing the ffmpeg folder with ffmpeg-master-latest-win64-gpl-shared.zip from https://github.com/BtbN/FFmpeg-Builds/releases renamed to ffmpeg
>>42992901same error again
>https://www.youtube.com/watch?v=WKFQdij7oXwthis random FS spanish rvc cover just came to my attention, so I thought somebody else would also appreciate it
>>42993208So it fails to establish a socket connection. Do you have an antivirus? Try disabling it, and run the .bat as admin.If that doesn't help, replace line 150 in PVT\runtime\Lib\site-packages\flet_runtime\sync_local_socket_connection.py with this like on picrel:import errno try: self.__connection, client_address = self.__sock.accept() except OSError as e: print(e.errno) print(errno.errorcode.get(e.errno))then run again. it should tell exactly what the issue is.I'll go sleep now.
>>42993360I dont have antivirus, so I just run with as admin, still giving me the same error.Debugging sure is fun
Up.
>>42993410>https://www.python.org/downloads/release/python-390/>The installer now also actively disallows installation on Windows 7. Python 3.9 is incompatible with this unsupported version of Windows.it seems like this may be the issue. 3.9 is not even a new version, it was released in 2020.You can confirm that socket communication is broken by activating the python env>runtime\python.exeand executing this>import socket;s=socket.socket(socket.AF_INET, socket.SOCK_STREAM);s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1);s.bind(('localhost', 8080));s.listen(1);print(s.fileno())if instead of a number it prints an error then it's windows 7
>>42994605this is a bit more complete>import socket;s=socket.socket(socket.AF_INET, socket.SOCK_STREAM);s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1);s.bind(('localhost', 8080));s.listen(1);print(s.fileno());s.accept()
>>42994605I do use Windows 7, however I am usually able to use any ai tool by installing conda env with python 3.10.4Anyhow, I've run the above and got the "204" number.Hmm, could you give me the exact pip freeze on what you have on your end, so to check if maybe there is something missing or some module is being pain in ass and needs downgrading on my end?
>>42994620also getting 204 with this one
>>42994622pip freeze is already in requirements.txtmaybe try with a new venv? but before that try reinstalling flet>pip uninstall flet>pip install flet==0.19.0
>Celestia sings There is nothing we can do Napoleon song>https://pomf2.lain.la/f/krdyl2ua.mp3I was trying to get this song >https://www.youtube.com/watch?v=F0Gkr4MBEO0 with Rarity or Fluttershy voice but the ai model was throwing too many weird pitch errors.
>>42994708 WARNING: The script flet.exe is installed in 'Q:\_AIfromC\_AItts\PoniVoiceTool\PVT\runtime\Scripts' which is not on PATH. Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location.ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.flet-cli 0.28.3 requires flet==0.28.3, but you have flet 0.19.0 which is incompatible.flet-cli 0.28.3 requires watchdog<5.0.0,>=4.0.0, but you have watchdog 3.0.0 which is incompatible.flet-desktop 0.28.3 requires flet==0.28.3, but you have flet 0.19.0 which is incompatible.flet-web 0.28.3 requires flet==0.28.3, but you have flet 0.19.0 which is incompatible.Successfully installed flet-0.19.0Due to python being dogshit at showing what dependencies are actually needed to install shit, I just force upgraded to the lowest version flet-cli was willing to accept the flet>runtime\python.exe -m pip install flet==0.25.2 watchdog==4.0.0 flet-cli flet-desktop flet-webI still get the message popup about >avutil-58.dll is missing>avutil-57.dll is missing>avutil-56.dll is missing>NFOutWorkItem2 could not be located in dynamic library kxdxSo I guess a part of the problem got fixed?
>>42994899>NFOutWorkItem2 could not be located in dynamic library kxdxkek, googling this currently shows multiple links to this thread and nothing else>avutil-58.dll is missing>avutil-57.dll is missing>avutil-56.dll is missingI have no idea what they are. try replacing the runtime with one included in "the integrated package" for windows here https://github.com/RVC-Boss/GPT-SoVITS which is where I got it originally. Then on top of that install app's requirements.
>>42992629This looks like an eldritch monster.
Pinkie Pie - Conga>https://pomf2.lain.la/f/u3gn2od.mp3
>>42972267new tts, wiphttps://files.catbox.moe/e8gnuu.mp3https://files.catbox.moe/v89nuo.mp3https://files.catbox.moe/wm0k98.mp3
>>42995291great song for ponka
>>42995502Will it stop sounding like ass when finished?
>>42995291nice!
>>42972267>>The PoneAI drive, an archive for AI pony voice content:>drive.google.com/drive/folders/1E21zJQWC5XVQWy2mt42bUiJ_XbqTJXCpIs this not getting updated anymore? I've never really looked at this general before but I wanted to listen to some random pony voice clips. I see that the newest stuff on here is from 2022.
>>42997350Yeah, who's even in charge of it?
>>42997731An anon. I used to manage it alongside said anon but stopped once content started slowing down.
>>42997874yeah, thanks for your work Anon. I would love to see it come back in some format, maybe an automated booru style with ability for the uploaders to put tags for characters and song styles and what not.
>>42997874So does any of this music get saved anywhere anymore?>>42995291>>42989243>>42977481And what about voice clips? Don't tell me you Anon's are creating this stuff and letting it die with the threads
Are we at the point we can have pony-narrated audiobooks yet?
>>42998940see >>42992585https://drive.google.com/drive/folders/16Ss26VBmgzcSuTGzhaHqRuyVRceTf-YBIf you can get this working, that's will be easy way. Otherwise there is the old way of just doing each line manually in any TTS and putting them together just like we've done five years ago.
>>42998982If you try it please report back whether you managed to launch it or not because I am not sure if I packaged it properly
>>42999237heyo, Im this Anon >>42994899 and I kind of had stupidly busy weekend + Monday, the soonest I can try to mess around with it will be Tuesday (im thinking of just setting up conda env on my end to see if that will be a good workaround)
>>42995502reee crapbox
>>42972267Hello all, I'm currently working on a theoretical project of a real-life rainbow dash. That's all I can reveal because the haters will sabotage me. Does anyone know if there's currently a solution for live, conversational talk with a Rainbow Dash voiced & personalitied chatbot that TTS's convincingly in realtime?
>>42999922uh might wanna look into local LLM's and TTSes?
Sorry, I don't want to be an annoying faggot but does can anyone answer my previous question? Is there a new place where pony voice clips and music is being uploaded to, if at all? I see that /create/ as this google spreadsheet for a bunch of songs but I don't know how regularly that one gets updated eitherhttps://docs.google.com/spreadsheets/d/1GrDEeCc9B7mAkQWAx_eoRdw4bF_W-5bsaAm4otvjfWE/edit?gid=0#gid=0
>>43000439>Is there a new place where pony voice clips and music is being uploaded to, if at all?No, no known active cataloguing efforts at this point in time.
>>43000443Damn. I guess I'll just save catbox and vocaroo links whenever I see something I like then. I don't know how many Anon's are still actively working within this thread, but thanks for everything you guys have done up till now
>>42995502having trouble getting these to load, anon
>>42999317I had some people with win10 test it and it works out of the box for them, so it's probably not a python env issue. Those avutil dlls belong to ffmpeg, the included version contains avutil-60.dll but yours reports absence of previous versions. I'd test whether the included one even runs and mess with path variables.
Anyone mess with MOVA at all yet? Wondering how it works for pony videos. The previous pony videos in this thread were so damn cool to see.>>42999922 (checked)>Making real life AI pony>That's all I can reveal because the haters will sabotage meCertainly won't get sabotaged here, you're in damn good company, kek. Unless you got some group of schizos on your tail, that I understand.
>>42977481Holy shit this post is blessed. I hope someone's backing these up. Does anyone have any more Pinkie covers? She's my fave, but truth be told I'll take any more in general because I love them. A Thousand Years by Fluttershy was so pretty. There's a wonder to some of these covers that's really touching me; combo of ponies and 2010's pop is a combo that sends you right back to 2012 and it's like being hugged by a warm blanket.
>>42999922>https://github.com/EDCD/EDDIAs a starting point I would recommend the github Elite Dangerous tts ai companion (it's one of few handful that are out there), it pretty much just something along the lines of connecting the SillyTavern with online LLM and TTS and some inputs from the game to have the chatbot acting as assistant/companion that chills out with you.>https://www.youtube.com/watch?v=n_m0fqp8xwQThere is also the new Nvidia idea PersonaPlex, were instead of having models pipe line of "speech to text -> LLM -> TTS" they kind have model that does all of the talking and listing at the same time, sadly its mostly trained on approved HR phone administration calls so the presented output is sucking ass at roleplaying characters.
>10
bump
>>43001298
>>42999922The only two that come to mind are:1. https://huggingface.co/nvidia/personaplex-7b-v12. https://github.com/FlashLabs-AI-Corp/FlashLabs-ChromaThe first one, an anon got it to run on his 3090 after some tweaking: https://desuarchive.org/g/thread/107886414/#107888720Result: https://vocaroo.com/1nLUHmvzYpUbNo idea how the second one performs.
>>43002522
