Description
Model + Dataset by me This is an update to my previous Jungkook RVC v1 model +Much higher pitch accuracy and realism +Used tensorboard to select model with minimal loss, before overtraining Settings: Crepe Hop Length: 16 Batch Size: 8 Epochs: 1175 Dataset Length: 6m 46s Kindly credit me if you use my model <
Comments
Samples:
Woah there This is definitely highly overtrained RVC v2 models with small datasets only need less than 200 epochs for good results
I checked the tensorboard, loss/g/total checks out
Here is the graph, the model is right before the 20k mark
Pretty sure you should have stopped at 12k
I could but the loss wouldn't be as low. I basically let it train for a long time and pick whatever point is lowest when I come back to the PC
not good. too soft
Good
Le puse un audio raro al modelo y lo canto muy afinado😭 pero si, se parece mucho a la voz de jk
Muy genial.
超喜欢
特别特别喜欢
特别特别喜欢,可以听到果唱我喜欢的歌
我要把我所有喜欢的歌,都让田柾国为我唱一遍,原来听懂你唱歌是这种感觉。
dear
太厉害了😭😭
good!
I made the voice of a demigod Chikn Nuggit!!!
Muy bien
awesome
好想哭
好贴啊
his voice is good for yellow
听的我想哭🥲
I love u E'zoza
Me encantó mucho ☺️
好喜欢他的语音
Se le salieron gallos al jungkook😭
我喜欢这个
Es bastante "Decente" Esta para usarlo en "ciertos temas" que no tengan nada que ver con BTS
谢谢大大做了这个模型,我太喜欢了
❤️
这个模型真的太可以了
หลับคาเกมดึกแล้วทำไมไม่นอนหลับดึกแล้วทำไมไม่นอนเล่น free fire ที่หลับสบายเมื่อไหร่จะตื่นมาฟังเพื่อนรักจนเพื่อนในทีมด่า
asdf
Add a comment
Samples
1. Singing
Male
English
2. Singing
Female
English
3. Singing (Dry)
Female
English
4. Singing (High)
Female
English
5. Singing 2
Male
English
6. Singing (Dry)
Male
English
7. Singing (Dry, High)
Male
English
Pitch
Weekly Metrics
Users also tried
More to explore
Loading more