SenseVoice deployment hub: llama.cpp, ONNX CPU, and private APIs

[LauraGPT]

Production deployment routes for SenseVoice

The new FunASR deployment hub collects the maintained SenseVoice deployment paths for local applications, CPU services, and private transcription APIs.

• Standalone desktop and edge packages: llama.cpp / GGUF guide with Linux, macOS, Windows, CPU, CUDA, and Vulkan release paths
• ONNX Runtime and C++ CPU services: CPU runtime guide
• OpenAI-compatible application integration: private API guide
• Reproducible container environments: Docker and Kubernetes guide
• Capacity, security, and rollback review: production checklist

The pages include model/runtime identity, commands, smoke tests, operations, security boundaries, limitations, and source evidence. Published benchmark records state incomplete hardware or timing information explicitly and require reproduction on target devices before a capacity decision.

See the canonical FunASR deployment-hub announcement for release verification, including static, browser, public smoke, TLS, and rollback checks.

中文和英文页面都已上线：中文 · English