Gawr Gura clean voice file

[Anonymous No.]

Anyone have an audio file of Gura's voice, with no background music, just her voice. I'm training a neural net to deepfake her voice so i can finally get her to said "I love you anon, please let me suck your dick UwU".

[Anonymous No.]

Basically any movie watchalong has no BGM.

[Anonymous No.]

Save us from this misery, Guinea Bee...

[Anonymous No.]

>>25714875
Any links? I just need a snippet of about 1 min of clean audio

[Anonymous No.]

>>25715037
https://youtu.be/_CEEbyjh4Wo
what ai are you using? I'm trying to do something similar?

[Anonymous No.]

>>25715310
Thanks for the link. I'm just testing out a bunch of different models like.
https://arxiv.org/pdf/1806.04558.pdf
https://arxiv.org/pdf/2005.05514.pdf

I'm gonna see which one mimic her voice the closest then i'll probably fine tune it from there. Who knows maybe i can get it light enough to do livedeepfake. I already have her 3d model, i just need to get an quest 2 before i can troll vrchat as Gura lmao
...

[Anonymous No.]

>>25714811
are you going to train tacotron 2 ?

[Anonymous No.]

who is the girl in the gura hoody?

[Anonymous No.]

>>25715556
sussh gura's previous identity

[Anonymous No.]

>>25715517
Definitely, i have heard some great results about that model. Also Tacotron's architecture is encoder-decoder which some thing im really interested in. Lmao my master's thesis is encoder-decoder

[Anonymous No.]

>>25714811
Make a hentai yo

[Anonymous No.]

>>25715757
Of course, why do you think i need her voice in the first place. I'm scrounging up enough cash for a Quest 2, i already have a stolen her 3d model. A VR hentai of Gura is gonna be so god damm fun

[Anonymous No.]

>>25715941
My penis and I pray for you. Hope the results will be worth the wait.

[Anonymous No.]

>>25716076
It's not gonna be finished anytime soon though. so dont keep your hopes up, im busy juggling between my job and my master's. Maybe by the end of this year if im lucky. But this is a project that will definitely be finished, i have poured way way too many effort for this shit. Heck im even using my school server's to train this shit

[Anonymous No.]

I don't know why you fucks get sexually aroused by Gura, nothing about her is boner worthy

[Anonymous No.]

>>25716913
I'm gay too

[Anonymous No.]

>>25716913
Loli is love loli is life. Flat chest is the best

[Anonymous No.]

>Janny leaves a rule-breaking thread up for 40+ minutes despite the report
Feedback on this shit moderation literally does nothing.

[Anonymous No.]

>>25714811
These threads are allowed now?
Can I finally make Senzawa jokes.

[Anonymous No.]

Wasn't there someone who literally already did this?

[Anonymous No.]

>>25714811
Can you make a vocaloid sound pack? onegaishimasu oppu.

[Anonymous No.]

>>25718064
No. They literally aren't. But the janny only responds to one kind of dox.

[Anonymous No.]

>>25718064
idk and idc, just pick a random pics of Gura. Seems like this one will be a nice reference. The main thing i'm here is for the links. And i already got it so the jannies can delete this thing idc

[Anonymous No.]

>>25715506
your coding them yourself? Holy shit good luck

[Anonymous No.]

>>25718115
Link plz, that would be nice because training this shit without any reference is time consuming asffffff

[Anonymous No.]

>>25718588
he spent months on it with hours and hours of training material and it was awful, robotic-sounding garbage, and he stopped posting

[Anonymous No.]

>>25718582
Not exactly by myself. Im using research paper's publish by organization like OpenAI and Microsoft. It's a very very bad idea to build a neural network from scratch. It is possible to do, but it will require a large team of expert ML scientist and the resources of a tech giant to do anything better than existing models

[Anonymous No.]

>>25718643
Do you remember which kind of models did he use or at least when he stop posting? So i can avoid using the models that he tried

[Anonymous No.]

>>25718799
i wouldn't tell you if i did. you should absolutely kill yourself and do it right now.

[Anonymous No.]

>>25718827
:(.
https://www.youtube.com/watch?v=PIh2xe4jnpk

[Anonymous No.]

>>25715506
How do you even start from this? I imagine stuff starts from a project from GitHub, compiling the source on your local and throwing the file path containing all the samplis on the console then the magic happens. Or in this case a console window showing an ASCII progress bar.

The hardest part I think is coding how the take the input, like the text "I love you anon" and how to handle the output, like binding the AI to your audio driver and trying to make a sound of Gura's "I love you anon".

[Anonymous No.]

>>25714811
>>25715506
holy shit, it's a great idea! Could you maybe make her sing a song? It'd be hilarious to hear Gura sing this shit https://www.youtube.com/watch?v=rlkSMp7iz6c

[Anonymous No.]

>>25719390
Yeah most of the time i read the research paper on arxiv to understand the process if there are implementations i will run in to see if i can replicate the results if it is within a margin of error like 2 to 3%(in case of classification problem) then i will usually give that model a go. For most problems i usually, run about 5 models, after that i pick the best one and then fine tune it for my specific problem

[Anonymous No.]

>>25719452
yeah i will try, but just keep the expectation low, cause even with the state of the art models the voice is still gonna sound like a robot. Im thinking of adding autotune to hide the robotic voice