Gordon Ramsay

# Before Start > - There are two versions of this model: the Ov2 Super version, and the V2 Simple version. The choice is yours! > - Don't forget that : the Ov2 Super version is still in the testing phase! > - Maybe you should do some tests with a microphone and a real voice (something I can't really do as I don't speak native English ...) because I still have the impression that the tts has a negative impact on the rendering (if I compare with the initial quality of my dataset). > - I find the Ov2 Super version super cool for 250 epochs (there are just a few little problems with its breathing, I think), while the V2 version seems a little less good for its number of epochs, but doesn't have the problem that I find more present with the other breathing, but with little parasitic noises. > - The dataset comes from raw audio files, belonging to a mobile game where he was able to lend his voice, I of course deleted all the files that could ruin the dataset (even if it doesn't seem to be enough ahah!). > - This is more of a "test" model where I see what Ov2 Super does, the V2 model is an old one I made maybe 2-3 weeks ago and forgot to post. > - GPT has played the role of Gordon a little too much is the way he talks, the normal v2 isn't that bad <:kekw:1010171252746502154> # Last Update : <t:1704852899:R> > - **Model URL :** > - *Ov2 Super Version *: https://huggingface.co/rayzox57/GordonRamsay_RVC/resolve/main/GordonRamsay_v2_ov2_250e.zip > - *V2 (48k) Regular Version *: **coming soon (i wait weights.gg corretly upload on website to avoid conflicts)** > - **Version :** > - <a:firev2:1167361499149381662> RVC V2.0 [ Ov2 / and / Regular 48k ] > - **Pitch Extraction Algorithm :** > - RMVPE > - **Epochs - Steps :** > - *Ov2 *: 250 - 12k (batch size : 4) > - *V2 Regular *: 350 - 8.75k (batch size : 6~8 i don't remember) > - **Dataset :** > - ~ 00:10:17 [ best voicelines of the mobile game: Gordon Ramsay's Chef Blast ] > - **Recommended Usage :** > - Speech > - **Search Feature Ratio :** > - 0.75 > - **Pitch :** > - Logic Pitch ( -4 / 0 = Man ) > - You can adjust if you find better version # Previews : > - **Preview_TTS_Ov2.wav :** > - *For* : Ov2 Super Model > - *Provide* : ElevenLabs v1 - Daniel EN > - *Script* : Generated by GPT 4 > - *Contains External Effects* : Yes > - *Pitch* : 0 > - *Feature Ratio* : 0.75 > > - **Preview_Cover_Ov2.wav :** > - *For* : Ov2 Super Model > - *Contains External Effects* : Yes > - *Pitch* : -4 > - *Feature Ratio* : 0.75 > > - **Preview_TTS_v2.wav :** > - *For* : V2 (48k) Regular Model > - *Provide* : ElevenLabs v1 - Daniel EN > - *Script* : Generated by GPT 4 > - *Contains External Effects* : Yes > - *Pitch* : 0 > - *Feature Ratio* : 0.75 >  > - **Preview_Cover_v2.wav :** > - *For* : V2 (48k) Regular Model > - *Contains External Effects* : Yes > - *Pitch* : -4 > - *Feature Ratio* : 0.75 > - *Additional info*: As you can hear, the cover rendering with v2 is more pitched than with ov2! So consider lowering it if necessary!

this deserves more love

^

Good job

<a:ShokoJoy:763196788151681036>