search

A mature Xiaoman RVC model with strong emotional connection

visibility 324chat_bubble_outline 0favorite 9
Updated: 26/07/14Published: 26/07/14
A mature Xiaoman RVC model with strong emotional connection

Inference Comparison

After Inference

0:000:00

Other Inference

0:000:00

Model Introduction

Before use, you can try listening to the sound.

This model is relatively stable in expressing emotions.

It is recommended to set the pitch to around 10 (adjust according to your original input timbre).

This model limits the enhancement of output volume when processing data; please cooperate with rack software to maximize output volume.

Training Log

g/total

tag: loss/g/total
klrvc.com212631364102k4k6k8k10k

Training Log Analysis

Training Name
mygf-xm
Sampling Rate
40k
Training Status
Finished
Training Epochs
300 / 300
Training Steps
8.3k
Save Interval
Every 30 epoch
Saved Epochs
30, 60, 90, 120, 150, 180, 210, 240, 270, 300
Recommended Best Checkpoint
epoch 92 / step 2.5k / loss/g/total 12.71
The green marker indicates the lowest loss/g/total point in this log and can be used as the recommended best checkpoint.
zhangzhang

zhangzhang

inventory_20person_add0

Version Details

OriginalYes
AlgorithmRmvpe
CapabilitySupport
LanguageChina, the US, Japan, etc
Sample Rate44K
Base ModelRVC_V2
HubertEnglish Hubert

Recommended Parameters

Threshold-60
Pitch10
Index Rate0.00
Volume Factor0.86
Sample Length0.30
Harvest Processes2
Fade Length0.12
Extra Inference Time2.92
infoFor reference only, adjust according to your input!

Related Models

No related models

Keywords

#rvc#Anchor#Emotions#Xiaoman

Comments

No comments yet