Description
Who: Navi is a recurring character in The Legend of Zelda series. She is a Fairy who serves as Link's companion in his quest to save Hyrule, often with annoying voicelines. "*Hey, listen!*" Stats - Trained on ~2 seconds of 16kHz ripped Voicelines from Ocarina of Time - Upsampled to 48kHz using audacity, and repeated 20x times to about 40 seconds of audio for processing purposes - 650 Epochs - RVC V2, local - Mangio-Crepe Hop Length: 64 - Never really finished training, but I retold the story of how I failed to make this model so I figured I'd post a revamp version \*Note: When inferencing, settings are everything for this voice. Try not to overuse the index (I used 0.8 for these samples) and make sure to get a generous mix of the original and input sound audio (I generated the inference samples with only a 0.6 envelope or lower). Also, make sure to pitch it up high, usually at least 12 for male input samples.
Comments

lmao bro went balls deep with this one

u shouldve done it with rmvpe

it would have a crispier sound to it

ohh yeah this ones a squeaker

all of this bc of me lmao

batch?

how did it take only 10 with a batch of

4

i mean 4 is kinda overkill here cuz the quality is poor no?

it's not a clear dataset

wouldnt like 16 be more appropirate?


why does kalo say otherwise?
How Did bro make this
Add a comment
Samples
Pitch
More models by