Simple stt(sound-to-text) locally
v1.0.1Simple local Speech-To-Text using Whisper. One-command install with auto model download. Supports 99+ languages.
⭐ 1· 404·2 current·3 all-time
byKuikui@lkisme
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description match the files and instructions: install.sh creates a virtualenv, installs openai-whisper and ffmpeg, triggers model download, and stt_simple.py transcribes audio and writes outputs. There are no unrelated credentials, binaries, or services requested.
Instruction Scope
SKILL.md and install.sh confine actions to installing dependencies, downloading the Whisper model, creating an output directory, and verifying installation. Note: the install script may call apt-get/yum/brew to install ffmpeg (system package manager operations) and writes under /root/.openclaw and ~/.cache/whisper, so it may require elevated privileges or path edits on non-root systems.
Install Mechanism
No remote arbitrary archives or shortened URLs; dependencies are installed via pip (openai-whisper) and system package managers for ffmpeg. The model is downloaded at runtime via whisper.load_model (network download, ~244MB for 'small'). This is expected for a local Whisper setup.
Credentials
The skill requests no environment variables or secrets. It only creates a venv and writes model/cache and output files to local paths. No extraneous credential access is present.
Persistence & Privilege
The skill does not request 'always' or attempt to modify other skills. It creates persistent files (venv, model cache, output) under /root/.openclaw and ~/.cache/whisper; installing system packages (apt-get/yum/brew) may require sudo. Consider adjusting paths if you cannot or do not want files under /root.
Assessment
This skill appears to be what it says: a local Whisper-based STT installer + small runtime. Before installing: 1) Inspect install.sh (already included) and be comfortable with it running system package installs (apt-get/yum/brew) and pip installs; those commands can require sudo. 2) Expect a network download of the Whisper model (~244MB for 'small') and increased disk usage in ~/.cache/whisper and the output directory. 3) Consider running the install inside a container, VM, or non-root user path (change VENV_DIR/OUTPUT_DIR) if you prefer isolation. 4) If you need to audit dependencies, verify the 'openai-whisper' PyPI package/source. 5) No credentials or unexpected external endpoints were found, but pip installs execute package code — only run if you trust the sources or run in an isolated environment.Like a lobster shell, security has layers — review code before you run it.
latestvk97802s6z65nvz9pqekgn0bwnx82jfmv
License
MIT-0
Free to use, modify, and redistribute. No attribution required.