search
Voice conversion assets

RVC Base Model
Download Center

Our models are continuously maintained. If you encounter any issues during deployment or fine-tuning, feel free to check the docs or contact our expert team.

RVC Base Model Download Center: This model is officially pre-trained and maintained by MiaoYin and is not authorized for distribution on any third-party platform. Our authorization system has recently revoked access for certain accounts that violated the rules. Please obtain models and updates only through official channels; versions reposted, resold, or misappropriated elsewhere do not include after-sales service, technical support, or future updates.
Fine-tuned / Fine-tuned

Mygf-f040K

A deep-optimized fine-tuned base for RVC v2.0, significantly improving capture of complex accents and specific timbres.

Sample Rate
40k Hz
Model License
MIT License
Recommended Duration
~30h corpus
File Size
~2.1 GB
Preserves original model stability
Optimizes dialect and accent details
Pre-trained / Pre-trained
New

Mygf-f040k

A newly trained base for the RVC 40k workflow, built from around 100 hours of data across 283 speakers. It is designed for everyday voice conversion and fast fine-tuning, with specific improvements to stability, convergence speed, and small-sample results.

Sample Rate
40k Hz
Corpus License
MIT License
Training Volume
~100h / 283 speakers
Min Fine-Tuning
From 7 minutes
Optimized for the 40k inference pipeline and common RVC training workflows
Broader speaker coverage for more stable small-sample fine-tuning
Purchased and licensed separately from Mygf-f048k
Pre-trained / Pre-trained ยท HOT
Recommended

Mygf-f048k

A fully self-developed base architecture with 48K ultra-high-definition sampling, breaking free from legacy limitations for professional creation.

Sample Rate
48k Hz Pro
Multi-Speaker
308 voices
Training Volume
~100h pure corpus
Min Fine-Tuning
> 5 min start
48K ultra-high fidelity
Independent neural architecture
SUPPORT

Need Technical Support?

Our models are continuously maintained. If you encounter any issues during deployment or fine-tuning, feel free to check the docs or contact our expert team.