Khasi RVC Voices
This repository contains Retrieval-based Voice Conversion (RVC) models trained specifically to generate realistic Khasi speech. These models are designed for voice conversion, dubbing, speech synthesis research, and improving Khasi Text-to-Speech (TTS) systems.
The goal of this project is to help support and expand AI accessibility for the Khasi language through open research and community-driven development.
Included Models
Both models were trained to preserve natural Khasi pronunciation, tone, and vocal stability.
1. Young Adult Male
- Model Weight:
Young_Adult_Male.pth - Feature Index:
Young_Adult_Male.index
2. Young Adult Female
- Model Weight:
Young_Adult_Female.pth - Feature Index:
Young_Adult_Female.index
Use Cases
These models may be used for:
- Voice conversion
- AI dubbing and localization
- Speech style transfer
- Enhancing Text-to-Speech (TTS) pipelines
- Creating Khasi AI voice outputs from existing speech
- Research in low-resource language audio AI
- Educational and accessibility projects
Ethical Guidelines
Prohibited Usage
You must not use these models for:
- Impersonation or identity fraud
- Deepfake scams or misinformation
- Harassment, defamation, or malicious content
- Generating illegal, hateful, or deceptive material
- Voice cloning without proper consent
- Commercial resale of the models without permission
Users are responsible for complying with local laws, platform policies, and ethical AI practices.