innnky/so-vits-svcArchived
So Vits Svc
This project is an AI voice training framework and singing voice conversion tool. It uses VITS and SoftVC technologies to transform the timbre of singing and spoken audio recordings, allowing a user to change the vocal characteristics of a recording to match a specific target speaker.
The system provides a web-based voice converter interface for managing model checkpoints and performing timbre transformation and pitch shifting. It supports exporting trained models to the ONNX format for use in external interfaces and lightweight runtimes.
The framework covers the full production pipeline, including audio dataset preprocessing, spectral feature extraction, and custom voice model training. It includes capabilities for generating voice profiles from vocal recordings and adjusting output audio frequency through pitch-controllable synthesis.
Features
- VITS Synthesis Models - Uses VITS architectures combining variational autoencoders and transformers to generate high-fidelity audio waveforms.
- Singing Voice Conversions - Implements speech-to-speech transformations that change vocal timbre while preserving the original melody and lyrics of singing.
- AI Vocal Production - Processes raw audio recordings into synthetic singing voices for professional music production.
- Audio Dataset Preprocessing - Cleans and standardizes raw audio datasets through resampling and partitioning into training, validation, and test sets.
- Voice Conversion Training - Provides a complete pipeline for preprocessing audio datasets and training voice conversion models.
- Voice Synthesizer Training - Trains deep learning models to capture and replicate the unique vocal characteristics of specific target speakers.
- Pitch-Guided Synthesis - Implements generative synthesis guided by fundamental frequency predictors to allow precise control over output audio pitch.
- Voice Identity Conversions - Transforms the vocal characteristics of spoken audio to match a target speaker with adjustable pitch transposition.
- Singing Voice Conversion Systems - Integrates training and inference systems using VITS and SoftVC for professional singing voice conversion.
- Audio Spectral Features - Extracts frequency-domain characteristics, including mel-spectrograms and pitch contours, to prepare audio for model training.
- Timbre Morphing Tools - Transforms the sonic characteristics of source audio to match target speaker embeddings for timbre conversion.
- Audio Feature Extraction - Generates spectral and pitch features from raw vocal recordings to create training-ready datasets.
- Reference Profile Generation - Generates custom voice profiles from recordings and base models to improve the stability of voice training.
- Real-Time Voice Transformation - Ships a web interface for managing model checkpoints and performing real-time vocal timbre transformation.
- Web-Based Speech Inference UIs - Provides a browser interface for uploading voice samples and triggering audio processing pipelines via a backend API.
- Web-Based Model Management Interfaces - Provides a browser-based interface for managing model checkpoints and performing voice conversion tasks.
Star history
How this analysis was created: This summary and feature list are AI-generated from collected project material and can contain mistakes. Stars, license and language are imported from GitHub. Inclusion does not mean that we have tested or audited this project. Check the source documentation for any feature you depend on. Learn more on our About page.
