GSV-TTS-Lite and Genie-TTS
These are complementary tools serving different inference needs for GPT-SoVITS: GSV-TTS-Lite provides a lightweight Python inference engine optimized for real-time performance, while Genie-TTS focuses on ONNX model conversion and cross-platform inference compatibility, allowing users to choose between native and standardized deployment formats.
About GSV-TTS-Lite
chinokikiss/GSV-TTS-Lite
GSV-TTS-Lite A high-performance inference engine specifically designed for the GPT-SoVITS text-to-speech model.(few shot voice cloning)
Implements millisecond-level latency through deep optimization techniques including Flash Attention support and decoupled timbre-emotion control, achieving 3-4x speedup on consumer GPUs while halving VRAM requirements. Provides multiple inference modes (streaming token-level output, batch processing, voice conversion) with subtitle timestamp alignment, and ships as a PyPI package supporting CUDA, MPS (Apple Silicon), and CPU backends via Python SDK, REST API, and WebUI interfaces.
About Genie-TTS
High-Logic/Genie-TTS
GPT-SoVITS ONNX Inference Engine & Model Converter
Converts PyTorch GPT-SoVITS models to optimized ONNX format for CPU-first inference with ~1.1s first-token latency and minimal runtime footprint (~200MB). Provides Python API, FastAPI server integration, and pre-trained character models across Japanese, English, Chinese, and Korean with emotion/intonation cloning via reference audio.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work