README.md

---
license: openrail
title: Real-Time Korean Voice Cloning
sdk: gradio
emoji: 📈
colorFrom: yellow
colorTo: red
app_file: app.py
sdk_version: 3.17.1
pinned: false
---
** Temporarily suspended

# Configuration

`title`: _string_  
Display title for the Space

`emoji`: _string_  
Space emoji (emoji-only character allowed)

`colorFrom`: _string_  
Color for Thumbnail gradient (red, yellow, green, blue, indigo, purple, pink, gray)

`colorTo`: _string_  
Color for Thumbnail gradient (red, yellow, green, blue, indigo, purple, pink, gray)

`sdk`: _string_  
Can be either `gradio` or `streamlit`

`sdk_version` : _string_  
Only applicable for `streamlit` SDK.  
See [doc](https://hf.co/docs/hub/spaces) for more info on supported versions.

`app_file`: _string_  
Path to your main application file (which contains either `gradio` or `streamlit` Python code).  
Path is relative to the root of the repository.

`pinned`: _boolean_  
Whether the Space stays on top of your list.


# Real-Time Korean Voice Cloning
This repository is Korean version of sv2tts. The original model (which was developed by CorentinJ(https://github.com/CorentinJ/Real-Time-Voice-Cloning)) is based on English.
To implement Korean speech on the model, I refer to tail95(https://github.com/tail95/Voice-Cloning). 
I changed some codes to improve convenience in preprocessing(audio and text) and training. Also I converted tensorflow model to pytorch model and fixed some errors.

## References
- https://github.com/CorentinJ/Real-Time-Voice-Cloning
- https://github.com/tail95/Voice-Cloning
- https://medium.com/analytics-vidhya/the-intuition-behind-voice-cloning-with-5-seconds-of-audio-5989e9b2e042
- Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis (https://arxiv.org/abs/1806.04558)


## Used Dataset
- KSponspeech (https://aihub.or.kr/aidata/105)

Make sure that your datasets has text-audio pairs.