File size: 1,310 Bytes
b434211
 
 
 
 
 
 
a59518c
 
 
 
 
 
 
 
 
28e5b1d
 
a59518c
28e5b1d
4524e35
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
---
datasets:
- RVC_V2
language:
- en
- ru
pipeline_tag: audio-to-audio
---
![Icon_Speechtest.png](https://cdn-uploads.huggingface.co/production/uploads/660c05092d6ffe8d9f59d1ed/KPuH4fhqxMBrQJNQ8keUL.png)
Voice models trained using soundbanks / speechbanks from Team17's Worms videogame series.
"w3dRUS" stands for the Russian localization of Worms 3D with its own soundbanks.

Some training material may be manually cleaned from SFX and long silences. Worms 3D variant of Whoopsie got the biggest edits, for whatever reason it had a few clips in noticeably lower quality, those were removed.

Most models use high-pitched voices which may work for female song covers without pitch change, or male song covers with Pitch set to 12.

![Icon_Speech.png](https://cdn-uploads.huggingface.co/production/uploads/660c05092d6ffe8d9f59d1ed/Yrntj-0xRZl8Hz-zfXaUa.png)
I'm not responsible for any malicious use of these models.

![Icon_Single.png](https://cdn-uploads.huggingface.co/production/uploads/660c05092d6ffe8d9f59d1ed/TxWK5yV0RJQTRtPk0AbUz.png)
Special thanks to the people who made it possible to extract voice clips from files in strange formats, uploaded them for the public, and helped me figure out how to use [RVC v2 Disconnected](https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS).