This Lora was trained on audio clips of JAV actresses in various states of stimulation.
The dataset was captioned in English & transcribed in Japanese, examples below. All of the words spoken in the dataset are Japanese, I don't know what happens if you give her an English/non-Japanese dialogue.
Sample training captions & phrases from dataset:
jpnMoans, she breathes heavily
jpnMoans, she moans [loudly/quietly]
jpnMoans, she moans and says "気持ちいい"
jpnMoans, she moans and whispers "やばい、だめ"
jpnMoans, she moans quietly and breathes heavily
jpnMoans, she moans loudly and says "あ、いく！”
jpnMoans, she moans and breathes heavily, saying "気持ちいい"
jpnMoans, she whimpers and moans
jpnMoans, she moans loudly, saying "すごい、いきそう"
jpnMoans, she whimpers and moans quietly
jpnMoans, she whimpers and moans quietly, saying "あぁ、いっちゃいそうそれ、だめ”
jpnMoans, she breathes heavily and moans quietly, saying "いっちゃった”
jpnMoans, she breathes heavily and moans loudly, saying "気持ちいい、ヤバ”
jpnMoans, she moans loudly for an extended time
jpnMoans, she moans loudly and says "気持ちいい～"
jpnMoans, she moans loudly and says "いっちゃいそう、いく！"
jpnMoans, she moans and orgasms loudly, saying "そこそこそこ、あぁ、あっいく、いくいくいく"
jpnMoans, she breathes heavily and loudly
jpnMoans, she breathes heavily and loudly, saying "いきそう～"
jpnMoans, she moans sharply and loudlyPhrases to know if you can't read Japanese:
気持ちいい - kimochii
やばい、だめ - yabai, dame
あ、いく！ - a, iku!
すごい、いきそう - sugoi, ikisou
そこ気持ちいい - soko kimochii
あぁ、いっちゃいそうそれ、だめ - aa, icchaisou sore, dame
いっちゃった - icchatta
気持ちいい、ヤバ - kimochii, yaba
ヤバい - yabai
めっちゃ気持ちいい - metcha kimochii
あっ気持ちいい - a kimochii
気持ちいい～ - kimochii~
いっちゃいそう、いく！ - icchaisou, iku!
そこそこそこ、あぁ、あっいく、いくいくいく - soko soko soko, aa, a iku, iku iku iku
気持ちいいー！ - kimochii-!
いきそう～ - ikisou~Description
FAQ
Comments (22)
Aaawww kawaiiiiii, sugoi desu ne? iku.
Hey man, I like what I like xD
masterpiece
tokyo hot?
how did you train audio only?
For Minimax lora training I've been using Fizgig, and you just make a dataset with only audio files.
@TwoMoreLurker Very nice, Fizgig is my go-to lora trainer nowadays.
hats off 10/10
“I tested it myself — even lines in other languages work really well. The audio is super clean, a big step forward!
Awesome! Yeah I tried to avoid audio with male grunts or especially noticeable "plap"s.
直接连接到lora节点使用就行了嘛？ 不用再额外添加音频文件了是吧
@chitangdebaba838 Yes, you can use it just like a normal lora with the FL2VA model, just add a lora loader to the model right after it's loaded.
Before: Load Diffusion Model -> Basic Guider
After: Load Diffusion Model -> Lora Loader -> Basic Guider
secret hack:
1. use audio vae output only
2.set laten size to 32x32 px
3.set a 15s-15s-15s....loop
bang! you got the best nsfw audio maker
Hello, I would like to ask how to write tags for videos? I used to work on images, but I found that these tags don't seem to be suitable for video generation. Do we need to add extra symbols
Cool lora! What is the ideal strength for this lora?
1 works well if you're not using any other audio loras. I'm working on some character voice loras too, and with those I set the character voice strength to 1 and this lora 0.4-0.6 and have been getting good results.
角色语音？是女明星这些的嘛？
@chitangdebaba838 As you can see from my other account I mostly do Persona stuff, so I've been working on cloning the voices of the Persona girls. It's taking a bit of time and I haven't quite gotten the training perfected yet, plus I have to curate a completely new dataset for each of them (since my datasets up to now have just been images).
This LORA is amazing; it really changes the tone of voice of characters. I've tested it in other languages, and you just need to change the format of the dialogue, like this: <d>[your language] dialogue</d>.
However, there is a problem: are there any suggestions in overall_soundscape that are more suitable for this LORA?
Hello, I would like to ask how to write tags for videos? I used to work on images, but I found that these tags don't seem to be suitable for video generation. Do we need to add extra symbols
When I use this Lora with Motion Context, I always hear a popping sound in the last second, but when I use this Lora on its own, everything seems to work fine... I've tested it many times, and the popping sound only occurs when Motion Context and Lora are used together. Does anyone know what's causing this?