32,011 Verified Open Models
28,877 Multi-Pitch Audition Ready
100% Direct Source URLs (HF · Drive · MEGA)
Model ID #32795 · Canonical Slug /model/1GPwdPvgGig
RVC v2
· RMVPE
· 40 Epochs
p323
RVC v2
RMVPE
40 Epochs
Peak Quality 4.82 MOS
Added Sep 20, 2025
https://huggingface.co/lyery/models-vc/resolve/main/p323.zip%3Fdownload%3Dtrue?download=true
Original uploaded URL returned HTTP 404 (huggingface.co/lyery/models-vc/p323.zip ) — automatically resolved to verified archive mirror above.
Convert Audio with This Voice in EasyAIVoice
Carries this exact model URL and your selected pitch (0 st ) into the cloud studio.
Zero GPU Setup
🎵 AI Song Cover / Singing
Auto vocal stem separation + instrumental remix
🎙️ Speech, TTS & Voiceover
Convert uploaded voice clip or microphone audio
Audio Decision Loop · Sample Deck
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.
★ Sweet Spot: FEMALE -4 st (4.82 MOS)
Female Ref
-16
-12
-8
-4
0
+4
+8
+12
+16
Male Ref
-16
-12
-8
-4
0
+4
+8
+12
+16
Amber border = Peak NISQA Quality
Raw model title: p323 (40 epochs, rmvpe, vctk)
CLAP Semantic Analysis · Voice Accords
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.
Auditory Discovery · Close Alternatives
Voices That Sound Like p323
Close but not quite the right fit? Audition nearby candidates matched by vocal embedding and timbre.
FEMALE · +8 st
4.80 MOS
Direct
FEMALE · 0 st
5.09 MOS
HF
FEMALE · 0 st
4.21 MOS
Drive
FEMALE · 0 st
4.79 MOS
HF
FEMALE · +8 st
4.51 MOS
Drive
FEMALE · 0 st
4.79 MOS
HF
Model Documentation & Notes
About p323 (40 epochs, rmvpe, vctk)
dl link: https://huggingface.co/lyery/models-vc/resolve/main/p323.zip?download=true
female speaker found in the vctk dataset, she is the speaker 88 of the original pretrain
this voice and the whole vctk dataset is free to use (Creative Commons Attribution License) but you must credit CSTR, University of Edinburg
works best for speech (audiobooks) but also has singing capabilities
trained on 20 minutes of data and a batch size of 8
Provenance & Open Access
Catalog ID: #32795
Weights Host: Hugging Face
