Korra
- Mangio-Crepe - Batch 9 · 32 hop length
https://huggingface.co/ai-characters/AI-Voice-Models-by-AI_Characters/resolve/main/korra-12min_truncated-rvc2-40k-mangio32-batch9_e21_s1008_by-AI_Characters.zip
Audition Across 18 Pitch References
Click any semitone cell below to hear how this voice performs on female and male reference vocals and lock your preferred pitch.
Sonic Profile at Selected Pitch
Acoustic traits measured directly from the generated output sample. Switch pitches in the Sample Deck to see how timbre shifts.
Architecture: RVC v2 · Pitch Extractor: Mangio-Crepe · Training: 21 Epochs · Source Host: Hugging Face
Voices That Sound Like Korra
Close but not quite the right fit? Audition nearby candidates matched by vocal embedding and timbre.
About Korra - RVC2 - Mangio-Crepe (32 hop length) - Batch 9 - 21 epochs
<https://huggingface.co/ai-characters/AI-Voice-Models-by-AI_Characters/resolve/main/korra-12min_truncated-rvc2-40k-mangio32-batch9_e21_s1008_by-AI_Characters.zip> Model was made by me (AI_Characters). Please @ my Twitter or Instagram accounts when using my model. I would love to know when it is used! If you want to support what I am doing, donations to my Ko-Fi (https://ko-fi.com/aicharacters) would be greatly appreciated! Those will go towards funding the renting of GPUs for further experimentation and model making. Dataset is from the entirety of season 3 of Legend of Korra, around 23 minutes of audio. With editing and silence-truncating it is down to 12 minutes of pure audio of her. Does very well for speech, and okay for songs. Some songs work better than others. She does not ever sing or do a high pitched scream in the show and her voice is naturally a little bit deeper for a woman so keep that in mind. I recommend keeping search feature ratio at or close to 0 for singing. Inference settings used for these samples, except for 0.5 feature search ratio for the speech sample: Mangio-Crepe, 64 hop length, 3 median filter, 0.2 volume envelope, 0.1 protect Note that the sample song vocals sound bad in the last minute. That's because the song has a lot of echo/harmonies at that part of the song that I was unable to remove. This sample is more of a hackjob than a true showcase of what the model can do, but I didn't have a better song that fits more to her voice on hand at this moment. Any helpful and honest critique of the model is greatly appreciated!