README.md

August 1, 2026 · View on GitHub

Description

This is an OpenVoiceOS speech-to-text (STT) plugin for whisper.cpp, a C/C++ port of OpenAI's Whisper automatic speech recognition model.

whisper.cpp runs on the CPU and has these traits:

Plain C/C++ implementation with no dependencies
Native support for Apple silicon, optimized with Arm Neon and the Accelerate framework
AVX intrinsics support for x86 processors
Mixed F16 / F32 precision
Low memory use through Flash Attention and Flash Forward
No memory allocation at runtime

pip install ovos-stt-plugin-whispercpp

The available models are tiny.en, tiny, base.en, base, small.en, small, medium.en, medium, and large.

  "stt": {
    "module": "ovos-stt-plugin-whispercpp",
    "ovos-stt-plugin-whispercpp": {
        "model": "tiny"
    }
  }

The plugin downloads models automatically to ~/.local/share/pywhispercpp/models/{model_name} on load.

This table shows the disk space and memory each model uses.

OpenVoiceOS/ovos-plugin-manager, which loads and manages OVOS plugins, including this one
ggerganov/whisper.cpp, the inference engine this plugin wraps
openai/whisper, the original Whisper model

Apache-2.0