Itaila V1.0 | RVC V2 RMVPE | 69 Epochs

[desert mesa]

🎙️ Italian Pretraining

This pretraining is beneficial for both speaking 🗣️ and singing 🎵 models in Italian.

📊 Training Information

⏳ Dataset Duration

The dataset of the model consist of 10 hours.

🔄 Epochs

The training involved 69 epochs.

💻 Hardware

Training was executed on an A100 GPU with 80GB memory.

📦 Batch Size

A batch size of 16 was used during the training.

🎚️ Sample Rate

The sample rate during the training was 32k.

🎼 Pitch Extraction

The pitch extraction was performed using RMVPE.

🎤 Voices

The total number of voices used for the training was 18.

⬇️ Download Link: https://huggingface.co/TheStinger/itaila/tree/main

[stuck girder]

programs won't download without .zip and ?download=true at the end

[solar inlet]

Heheheh... Nice

[stuck girder]

please fix the link

[rigid crag]

Add pitch extraction too

And u are set

[desert mesa]

done

[desert mesa]

what programs

[rigid crag]

Slay ✨✨

[reef grotto]

Add the dataset duration, AKA how many minutes or hours of speech the model's been trained on

[desert mesa]

it says 10 hours

[reef grotto]

That doesn't say anything about the dataset, though

[desert mesa]

fixed

bad gpt moment

it changed what i asked

thanks for the heads up

[reef grotto]

Thanks

[stuck girder]

Applio

[desert mesa]

this is not a model

[stuck girder]

What exactly is this?

[desert mesa]

youre an helper.........................................................................

[stuck girder]

dataset?

[rough lance]

it's a pretrain which you have to manually download and put in your program and then use it instead of the original v2 pretrains

[reef grotto]

Pretrain

[real vessel]

10 saatlik dataseti nerene sokcan

[stuck girder]

dataset mi demek istediin diyordum

dataset napayım ben

[real vessel]

Doğru

[south garden]

Ayo? @real vessel level 13 !!!

[deep nebula]

69 Epochs