PriesiaMioShirakana/DragonianVoice logo

PriesiaMioShirakana/DragonianVoice

Free

多个SVC/TTS的C++推理库

FreeFree tier
Inputs: text, audioOutputs: audio
Type
Open Source

About PriesiaMioShirakana/DragonianVoice

DragonianVoice is an open-source C++ inference library that provides a unified ONNX runtime for multiple speech synthesis and voice conversion models. It supports a wide range of TTS (Tacotron2, Vits, EmotionalVits, BERTVits2, GPtSoVits), SVC (SoVitsSvc v2/v3/v4, RVC, DiffusionSvc v1/v2, FishDiffusion, ReflowSvc), and SVS (DiffSinger) architectures. The library offers C/C++ and C# APIs for easy integration and now operates as a library-only project after discontinuing its GUI. It also integrates fish-speech via the ggml framework. DragonianVoice runs completely offline, requires no Python environment, and converts models to ONNX format for efficient inference.

Key Features

Supports TTS models: Tacotron2, Vits, EmotionalVits, BERTVits2, GPtSoVits
Supports SVC models: SoVitsSvc (v2/v3/v4), RVC, DiffusionSvc (v1/v2), FishDiffusion, ReflowSvc
Supports SVS model: DiffSinger
Uses ONNX runtime for cross-platform inference
Provides C/C++ and C# APIs for easy integration
Integrates fish-speech using ggml framework
Runs completely offline without Python environment
Fully open-source and free

Pros & Cons

Pros
  • Free and open-source with permissive license
  • Supports a wide range of popular voice models in one library
  • Cross-language API support (C/C++/C#)
  • No Python or deep learning framework setup required
  • Operates fully offline, privacy-preserving
  • Active community with QQ group for support
Cons
  • Requires conversion of models to ONNX format before use
  • GUI maintenance discontinued; now a library-only project
  • Some DML issues with older SoVits versions
  • Documentation primarily in Chinese, limited English support
  • No built-in model training capabilities

Best For

Text-to-speech generation for creative projectsSinging voice conversion and synthesisVoice synthesis for game modding (with restrictions)Research and experimentation with voice modelsLocal deployment of voice AI without cloud dependencies

FAQ

这个项目到底有什么作用?
项目的开发初衷主要是实现无需环境部属各个语音合成项目,现在打算制作为一个SVC的辅助编辑器。它简化了语音合成项目的部署,支持多种模型。