runtime-llamacpp-v0.1.1

diffusionstudio/coreruntime-llamacpp-v0.1.1Jun 21, 2026by LauraGPT

AI Summary

This release introduces prebuilt, self-contained binaries for the FunASR llama.cpp runtime, enabling local execution of SenseVoice (and Paraformer / Fun-ASR-Nano) speech recognition models. It features built-in VAD and strong performance on Chinese and Cantonese without requiring Python, build tools, or PyTorch.

Key Highlights

Prebuilt binaries for Linux (x64/arm64), macOS (arm64), and Windows (x64)
Support for SenseVoice, Paraformer, and Fun-ASR-Nano with strong performance on Chinese & Cantonese
Built-in FSMN-VAD for voice activity detection
Zero-dependency execution (no Python, torch, or build required)
Improved accuracy: SenseVoice 8.01% CER vs whisper.cpp small 22%

New Features

SenseVoice ASR model support
FSMN-VAD integration
Cross-platform binary distribution
Zero-dependency architecture

Full Release Notes

Prebuilt, self-contained binaries to run **SenseVoice** (and Paraformer / Fun-ASR-Nano) locally with the FunASR llama.cpp / GGUF runtime — built-in FSMN-VAD, whisper.cpp-style on-device ASR, strong on Chinese & Cantonese.

```bash
bash download-funasr-model.sh sensevoice ./gguf
llama-funasr-sensevoice -m ./gguf/SenseVoiceSmall-f16.gguf --vad ./gguf/fsmn-vad.gguf -a audio.wav
```

No Python, no build, no torch. Binaries for Linux (x64/arm64), macOS (arm64), Windows (x64). Docs: `runtime/llama.cpp/README.md`. CER (micro-avg): SenseVoice 8.01% vs whisper.cpp small 22%.