runtime-llamacpp-v0.1.7

modelscope/FunASRruntime-llamacpp-v0.1.7Jul 18, 2026by github-actions[bot]

AI Summary

Releases prebuilt self-contained binaries for the FunASR llama.cpp runtime, supporting SenseVoice, Paraformer, and Fun-ASR-Nano with built-in FSMN-VAD, including specific assets for CPU (x64/AVX2) and Windows CUDA.

Key Highlights

  • Prebuilt self-contained binaries for llama.cpp
  • Support for SenseVoice, Paraformer, and Fun-ASR-Nano with FSMN-VAD
  • Includes x64 and x64-avx2 CPU assets
  • Includes Windows CUDA asset (CUDA 86 architecture)
  • No Python ASR runtime or local build required

New Features

  • Self-contained llama.cpp runtime binaries
  • Windows CUDA runtime support
  • Built-in FSMN-VAD integration

Full Release Notes

Prebuilt self-contained binaries for the FunASR llama.cpp / GGUF runtime: SenseVoice, Paraformer and Fun-ASR-Nano with built-in FSMN-VAD. Download the default quantized model with `bash download-funasr-model.sh <sensevoice|paraformer|nano>` (the helper requires the Hugging Face CLI: `pip install -U huggingface_hub`), then run `llama-funasr-cli` / `llama-funasr-sensevoice` / `llama-funasr-paraformer`. Use the default x64 asset for maximum CPU compatibility; use the x64-avx2 asset on CPUs with AVX2/FMA/F16C/BMI2 for higher throughput. The Windows CUDA asset is `windows-x64-cuda`; it requires an NVIDIA driver compatible with the CUDA Toolkit version configured by the release workflow, targets CUDA architecture 86, and enables SenseVoiceSmall graph execution with `llama-funasr-sensevoice ... --backend cuda`. Build from source for other GPU architectures. No Python ASR runtime or local build is required. Docs: https://github.com/modelscope/FunASR/blob/runtime-llamacpp-v0.1.7/runtime/llama.cpp/README.md