Voice actors on suicide watch once again; converts spoken content to other languages in minutes, while preserving the tone of the original voice(s).
https://twitter.com/elevenlabsio/status/1711739680419414411
https://elevenlabs.io/dubbing (No email or login required)
Post results.
>dubs
Not all of us are neets
Is this the only retort you fags have? It doesn't even make sense here.
Imahine, because we all k ow you don't
> work all day
> wife and kids
> socialise
After all that, you do not want to learn the language of a coubtry that can't win a war or read undertitles.
Do you have trouble reading books too anon?
I'm surprised you can read this post.
>zero motivation to do anything but meet his bare minimum social obligations
>calls me a neet
> work all day
> wife and kids
> socialise
lmao you do this? why? for what purpose?
being productive for your boss, paying taxes, making sure the elite have a new generation of golems to manipulate, and engage in peer-to-peer brainwashing?
you're insane
Thank you for blessing us with a piece of your very limited daily freetime allowance, homosexual.
>work all day
Gotta put food on the table
>wife and kids
Got the first, working on the second, is pretty great when your girl shares your interests
>socialize
Dear God please no more I don't want to be perceived anymore.
Imagine WANTING to read? Fuck that
It’s going to sound like the same VA but speaking another language
my wife chino says it's okay to dub english media to japanese
only valid criticism of dubs are them not keeping the energy of the original voices
wrong, the real problem is that japanese writing is cringe as fuck and sounds like a 14 year old's power fantasy, so not having to hear it out loud helps
My biggest problem is that it's not uncommon for voice actors to change between seasons.
Especially with multimedia franchises.
this. who gives a fuck, make my chatbots feel more alive
anyone experiencing higher pitch outputs with updated tortoise?
just upgraded from vanilla tortoise and all the outputs are pitched up, currently
manually pitch shifting the audio file down -4% to counter the effect
your sample rates might be fucked somewhere
44100 or 22050 recommended?
i just did a test with 22050
output was worse quality with the exact same settings
voice was pitched up
any help on this
>22khz sped up to 44khz sounds low quality and/or pitched up
wow, this is an anomaly, i wonder why this happens, guess we'll never know
origianl was with 44k, same chipmunk effect
Honestly, with how cheap it is for the paid plans, this is a massive deal. Especially for more niche languages. Expect this to be used for lesser used languages since nobody would ever be called out on it.
The Star Trek universal translator™ is getting closer every day. Maybe one day everyone will be speaking the superior language: English.
if you truly believe that then you are brain damaged.
> that low sample quality
> that shitty synth speech
wow. we really are living in a magical world of star trek-like technology!1 coming up next: transporters! two more weeks, guys!
no, seriously: you're immensely retarded and entertained like a monkey being shown magic tricks. let me know when this tech isn't shit.
Are you oblivious to the fact that this is one of the fastest advancing fields of technology today? It's not "going to happen", there will be no great breakthrough and it will rocket like this until it will be able to sate most people and most people don't have very high standards.
Don't bother people with that mentality. It's the next step for luddites. Instead of just hating technology blindly it's just endless goalpost moving to look more reasonable. They did the same thing with computers and phones saying they are too heavy to ever be useful. Early Guns and automobiles had the same type of critics
Problem is that third worlders genuinely can't speak well, nor can they even write their own language properly. You'll notice this if you play MMOs.
I honestly see the advent of true working universal translators as the death of English language domination
i love having liz wheeler read harry potter to me
HeyGen
Has translation, dubbing, and video lipsync correction
>wants my email
How about... no?
>not having fake emails to dump junk to
isn't heygen just run through elevenlabs?
Dunno, but its much more accurate imo
blob:https://elevenlabs.io/efd4d0cb-73f6-49f6-b574-70075ca4562d
Yeah works great, not a nightmare at all
rarted
It's a massive pity no one has been able to make an open source TTS that is on pair with this.
We can clone voices with open source to a decent quality, but TTS voice cloning is still robotic as fuck and 90's-tier.
Because the space only has
>corponaggers nickel and diming shit
>ethicist researchers keeping the shit to themselves
>nodev chinks with lab-tier hardware producing slop
>the ick on eck fat hairy homosexual that refuses to accept hardware only to slowly train a model, just to restart it again
>rehashed slop trained on too little data
You're forgetting that for voice-shit to actually prosper like SD/LLAMA, you'll need both a corposhit model leak AND a dev/community of devs to work on features and whatnot around said model.
The latter is crucial, as between tortoise => sovits => bark => rvc => vall-e x, the ONLY ones that have actually had any attention were the ones that have had someone carrying development (tortoise with ecker and sovits/rvc with the circle of asians, ignoring their repo dramas).
I just want my waifu to read me pages and books, which of those will help me accomplish that the best?
closest it training your own version of matcha tts using a specific dataset of the voice you want
https://github.com/shivammehta25/Matcha-TTS/issues/9
Japanese chainsmoker Vegeta.
look at what ((they)) need to do to mimic a fraction of her power
Down syndrome Vegeta with a British accent.
sounds like a drunk japanese man singing karaoke
Anyone getting
>Watermark not allowed for audio input.
?
upload video, not audio
Thanks it worked.
Original:
Crazy one. Lol at the random "subscribe to the channel"
This is unironically good, LMAO.
Jap Rap doesn't so bad
HIROHITO SURRENDER BROADCAST 1945
Kek
Original:
This sounds like something from Assassin's Creed 1.
song name?
AAH SEBULBA NUMBER ONE EH!
>subscribe to pewdiepie
What did he mean by this?
Eleven labs is racist against indians it seems.
HOIST THE FLAG OF COMMORAGHHHHHHHH
lmao
>autistic screaming
why am i laughing at this
Screams just don't work for some reason.
Laughing and other non-word sounds makes it produce weird noises as well. Lyric-less music too.
ILL DIRECTLY BEAT YOU UP
Lmao what
In both and
it hallucinated some standard youtuber speech, which leads me to think they just analyzed a shitton of videos with cc enabled and are basically using a variation of their TTS model.
>basically using a variation of their TTS model.
Was there any doubt?
Is he ok?
I was planning on not sleeping this week anyway
Original:
>Internal Server error. All such crashes are reported to us automatically.
10/10
>Watermark not allowed for audio input.
There's a whole untapped world of non-English memes out there
Holy shit
Could you post the original please?
Muchas gracias hermano.
Holy cow that
>ding
Dang
the best is the mariacha band doesnt miss a beat kek
True consummate professionals.
>BULL GOT HIS ASSHOLE
>BULL GOT HIS ASSHOLE
Ah, the recovering stroke patient edit.
CS:GO
>Stephen Hawking Plays CS:GO
>finally going to get my Operation Meteor and Witch from Mercury dubs
Probably not funny for non french but I can't stop laughing
>what a minute, what a minute, what is this
cursed
English to English cuts the IQ of the speaker by 50%.
behalve in nebraska
lol, what even is that accent it gave kazuma
>what even is that accent
GOOD MORNING SIRS MY NAME KAZUMA
How does it manage to make the dub of the girl so bubbly? Like the way she says "power" at the end with a really strong "p" just the way bubbly characters do. Incredible
On one hand it's total dogshit with bizarre voices, not even self-consistent sometimes, the timing is off, can't handle emotion well, completely just glitches out sometimes, etc.
But consider that that in a lot of obscure language markets that have little budget to dub foreign content you will literally see shit like 1 bored guy voicing all the characters including the women.
Now they can just go: insert English show -> consume 8 cents of electricity -> dubbed show and the little kids in Fuckistan who watch Fuckistani dubbed cartoons won't know the difference.
Which means it's already a marketable product even though they're really just advertising to get investment capital to make a 10x better version of this in a few years.
Or as likely just retire with a billion zlotys and let Google or Microsoft buy them out and make their version, where 80% of the time the audioclip you get is just a dog barking and at least one of the characters always sounds like an African American for some reason.
holy shit this is bad. their voice-to-text one was good with aqua
cursed as fuck
Yakuza In Dutch is awful
Yanny/Laurel solved once and for all
some anime clips from jap to polish
>frieren
>attack on titan
>jujutsu kaisen
it's pretty shit, the automatic translations are awful, it needs a good amount of dialogue to pick up the voice, and the free version has loads of watermarks in audio.
I remember talking about this in those early elevenlabs threads. This could actually be pretty good, dubs sometimes have inconsistencies when it comes to certain voice actors (as a random example, the va for Spike in the Latin American dub of Cowboy Bebop didn't reprise his role for the movie, being replaced with another guy), so this could be used to bring old voices back to shows/movies where some voice actors just aren't available anymore. I know this can already be done with regular ai, but aside from that I guess you can now (re)dub your favorite chinese cartoons to English (or any other language) using the original va in an easier way.
It's pretty good. Not perfect yet, but pretty good.
WTF this shit is schizophrenia simulator
https://streamable.com/xt5ha0
>This video has been removed for violating our Terms of Service.
wtf indeed
Finally I can understand my Chinese JOI videos
>do NOT let it come out!!!!!!
or kek at them at least
Oktavist English to English just keeps both. ALmost like it detects it as a voice for the TTS, but not for whatever removes it from the background.
Original:
It does the same with vocoded vocals
デベロッパー デベロッパー デベロッパー
I can't believe this coked up retard makes so much fucking money
Guess confidence is important to make it big
he's living the life we all wish we could have
It is a shame this does not use text files as a source. Many laughs could be had if i could right my own scripts. It is quite impressive tech, the dubbing is a little off in some places. But that might be a flaw in the source files.
that part got locked down
Yea, shame. Fuckers are anti-fun homosexuals.
Who the fuck cares if they make you say naggerhomosexuals as a public figure?
Well more than a little i guess. This one is a failure. It just makes a small noise at the beginning. Was supposed to be Jap. i assume it is trying to say Hai but it sounds definitely wrong. So it needs clear samples of speech with no background noise.
I wonder how it takes to TTS programs. Anyone try MS TTS's?
>he dubbing is a little off in some places
"A little".
Just had feedbacks speding months with my friends playing with pic related without speaking English or Spanish and generating so many inside jokes
Good days
holy shit I had completely forgotten about htat kek
>feedbacks of*
Fuck I still can't speak English.
Flashbacks*
Goddamn it.
the ancient site is somehow still up https://www.text2speech.com/
Is there a local model on par with elevenlabs yet? If not I'll ask again next month..
40k sounds gay in french.
This sounds like actual good lines for a WW1 game.
Try actual porn with moaning.
Bonus points for trannies to see if they get assigned a male or female voice
Oh finally, i can rewatch DBZ without cringe ass Grandma Goku and stupid ass Schemmel
Not so fast!
https://streamable.com/sclz0a
AI is a meme
Did it decide he was french because of the Piano? I have heard 11ai make english accents out of a japanese voice, but never a french one.
I'm trying to feed it loud nigra but it just fails after a while.
Does this mean we can make Japanese voices speak proper English?
Check the thread and find out for yourself
not really, it's mediocre at best
Can someone do these two?
Why can't you do it?
It keeps refusing everything I try.
It started with the loud nigra, maybe because of the title, or maybe because I made so many by restarting incognito sessions, but it seems like my (static) IP got banned or something.
https://streamable.com/1c7yrp
>I'm jaaaaaaccckkiiiiiing my meah
There is no way this is the original audio dude.
When it can't recognize language it hallucinates the shit out of it
Listen to these:
it probably got fucked up from trying to process 49 voices all layed on eachother
Sorry, catbox seems to be down.
Reuploaded here
https://streamable.com/0g980c
https://streamable.com/uzp266
https://streamable.com/5cfe1j
Thought the catbox issue was on my end. Topkek I'm dying thanks for the laughs!
Nvm, here's the rest of mine + some I've downloaded
https://streamable.com/2jh5ff
https://streamable.com/wq50jv
https://streamable.com/nd5bpn
https://streamable.com/1etmbm
https://streamable.com/gsm3w0
https://streamable.com/190248
https://streamable.com/j1wt7d
https://streamable.com/9zsr0q
https://streamable.com/awakta
https://streamable.com/5nz3sq
https://streamable.com/bif47z
https://streamable.com/9gvvjn
https://streamable.com/9kpgyk
https://streamable.com/1uhnp6
https://streamable.com/ccy1s7
https://streamable.com/ia156y
>sweating like a downie ass
Superlative
https://streamable.com/g8p853
Kek thanks
catbox wont load so i cant hear any of your shit
if you're fast enough you can get a new one in before it loads the old one
I can't even download these
https://streamable.com/qxw4e9
https://streamable.com/p7xad8
Who would have thought transformer models would be good for translation. What can't it do!
automatically scanlate my doujins along with the SFX which is always written messy and unrecognizable
>doujins
Leek Spin
https://streamable.com/kphg2a
lol
is it a good idea to microwave this japanese dub
not bad
redubbed back to english
how ESLs are made
it can't do singing lol
that one was russian, finnish big iron is even worse
What's the limit? Can it dub a long podcast?
hola!
Not bad that it can recognize improficient speech.
This thread sucks and you all need to go see otolaryngologist and fast.
It would be pretty good it stressed the words in right place when transferring the interpretation from the original audio.
Already a thing with local models.
Beautiful.
There will be people watching series exclusively like this in the future and then there'll be nostalgic fags saying this was peak anime.
Is 3 minutes the limit?
I tried a 4 minutes long clip and it only generated the first minute of it.
Happened to me too, if its any longer than a minute it shortens it to 58 seconds apparently.
I don't care what the luddites say, this is Art
Retarded Bot.info shilling for free again
>free thing
>shilling
not everyone is an anti fun homosexual like you
>elevenlabs
>free
what
Everything posted ITT was made for free
And it shows kek.
And?
this is great
Such a wise and noble samurai
>DOKU STUKKU! DOKU STUKKU! お願いします
kek
i kneel
>dicku sakka
Damn this is one of those days I really wish I didn't have to go to work. You fuckers will have mined all the keks and gotten the thing lobotomized out of fear before I get home tonight.
post some videos, maybe some anon will run them through it
doubt there's need to worry about them fucking with it, it's pretty much harmless
i think he means that after "aboose" they will lock it down and render it useless like bing before he can come back and have fun with it
that's cool, can I use it myself to translate anime?
english dubbed to russian, redubbed to japanese, redubbed again back to english https://files.catbox.moe/tv17cw.mp4
Fucking kek
>changes voice and accent
>"Oh my god! My little meat turned into a big mochi!"
>goes back to old voice
Grobaristos
doesn't have quite the same tone as the original: https://youtu.be/6nT-5pzV1Ps
Anime voice actors are a bunch of insufferable cunts. I hope the Jap companies steal their voices as intellectual property.
People standing in line to meet American anime voice actors is one of the most bizarre occurrences in western fandom. These are people who have zero relationship with the actual work.
I can kinda understand in the case of the DBZ voice actors, the Evangelion cast, steve blum, Kari wahlgren (basically, shows you almost certainly had to experience on tv and most likely dubbed first), but less so for the voice of any KnY character, or the latest CGDCT to get butchered.
Finally I can share my french videos with the world
Jk this is a mess
horse joined them
meanwhile open sores can't even clone a voice without training for 12 hours or jumping through 3 different neural nets
lmao
elevenlabs is literally using those open source tools but with more glue
>but with more glue
Which is?
And why can't open source tools use similar "glue"?
Yeah, open source just doesn't have a chance against the big players like 15 polacks in a basement. Let's just give up and accept that text to speech is censored for our own safety from now on.
putting it back to english gave it a spanish accent
Is there an open source alternative to this?
Never ever
Buy the elevenlabs
>the abortion is a success
>thanks rabbi
it is known
PEREGRINO DESCARADO
Did they train this on youtube ASMR videos or some shit? So much random creepy whispering in all of these.
english lagtrain
I honestly don't know why I expected it to be any good.
kino
I love AI so fucking much bros
Why not have an eye on a cutting-edge platform that integrates AI generators like OpenAI's GPT 4, GPT 3.5, GPT 3, Dall-E, Ada, Curie, and Davinci. From coding to chatbots, text to images, and even VoiceOver magic, AGII is your go-to for amplifying your Web3 projects with AI wizardry!
What happened to the Bot.info own elevenlabs fork?
It turns out from /tesg/ thread at /vg/ that all of the anti AI crusade pulled by that Robbie guy was a Psyop by the VA Union to demonize AI and the robbie guy was in fact the fat israelite who calls himself a voice actor on twitter and Gianni administrating the twitter and reddit accounts, hence why he got into the jounos ears so fast.
It's getting there kek
DUB: https://files.catbox.moe/msk24h.mp4
Original https://www.youtube.com/watch?v=sfWARF7DmM4
AI-dubbed Luvia is a goldmine of hilarity.
>it's actually a convincing shinji
I NEVER DOUBTED
SPIKE SPENCER I KNEEL
Could this be another AGII in disguise? An AI platform that leverages the power of advanced AI generation models and has now become a top challenger with the likes of ChatGPT, Midjourney, Dall-E, Cleverbot, etc., providing something more than just AI assistance really deserves some respect. Thanks to its vast array of tools
>for dubbing songs it just talks
so close. soon i'll be able to fill my autistic itch to hear songs i like in japanese.
>Voice actors on suicide watch once again
Absolutely retarded take. This will be YOUR voice soon enough
AI voice acting is going to be great for mods
for instance, you could make an entirely voiced side character for baldur's gate if you wanted
I just want an anime girl to berate me constantly, why do we not have this technology available yet?
It's possible now if you set up a chatbot with the personalty you want and pass it through to text to voice generator trained on an anime girl you like you can even make porn of them.
Wtf does this mean?
>Watermark not allowed for audio input.
It needs to be a video file.
I tried a youtube vid and then tiktok. It says dubbing and then stops. No output. I have used up the free tier?
can voice actors just hurry up and kill themselves already
Can anyone give this an English dub?
It won't let me dub anymore.
Fantastic, now I will finally be able to enjoy Star Trek as it was originally intended, in the original English.
SQUALARA
Absolutely scuffed. https://files.catbox.moe/f3ohyz.mp4
why these language shit never have vietnamese? damn god life ass
Shinji is pretty hoarse after going through it all.
fucking awful lmao
>killing thousands of jobs with every update
>little to no media heat
how?
>thousands of jobs
This is bait but there's a lot of stigma with using AI tools, not to mention the questioning of legality, license, and ethics of it.
Most importantly, most of those "thousands" are bottom rung people working the equivalent of McDonalds jobs so naturally their voices aren't going to be heard.
>little to no media heat
they're not paying for advertising like chatgpt did