search

RVC-Project / Retrieval-based-Voice-Conversion-WebUI

Easily train a good VC model with voice data <= 10 mins!
MIT License
24.18k stars 3.58k forks source link

how to combine voices ? #608

Closed KabetzuriBoii closed 6 months ago

KabetzuriBoii commented 1 year ago

i was told one could merge multiple voices into one and create a original new voice . No idea how to pull that off though .

Yurzi commented 1 year ago

you can try to use the ckpt processing tab in gradio webui

TripleKiller666 commented 1 year ago

Very simplified Explanation: Step 1:Train Model A and Model B Step 2:Go to "ckpt Processing" Step 3:Insert Full Path in "Path to Model A:", same for B (ex. "G:\RVC-Webui\weights\keanu_en.pth" Model A, "G:\RVC-Webui\weights\franksinatra_en.pth" Model B) Step 4: Choose how much of Model A you want to Dominate in the generated Model (Weight (w) for Model A) 0.5 means 50% of the Voice, and if you just want to mix 2 Voices, is generally good enough to leave as it is. You can still try out other combinations latter. Step 5: Don´t forget to give the new Model a Name, and set it to the same Version as the Models you trained before (so if both were v2, then this Model is v2 as well), mixing different Versions didn´t work for me, so either stick with v1+v1 or v2+v2. Step 6: Test the new Model on the Inference Tab!

github-actions[bot] commented 6 months ago

This issue was closed because it has been inactive for 15 days since being marked as stale.