← Back to Skills Marketplace
openclaw-whisper-voice
by
sabyaghosh
· GitHub ↗
· v1.0.0
· MIT-0
193
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install openclaw-whisper-voice
Description
Local Whisper speech-to-text for audio files and inbound voice notes on the OpenClaw Gateway host. Use when setting up local transcription for WhatsApp, Tele...
Usage Guidance
This skill appears coherent and implements local Whisper transcription as described, but review and consider the following before installing:
- The installer will pip-install a large set of Python packages into your home (~/.local) and will download model weights into ~/.cache/whisper on first run. Expect significant network, disk, and CPU usage.
- The install script fetches get-pip.py and uses pip from public indexes (PyPI and PyTorch wheel index). If you require stricter controls, run the install in a virtualenv, container, or on a throwaway host; or mirror/verify packages internally.
- The scripts create ~/.local/bin/whisper and a symlinked ffmpeg shim. If you already have system binaries with the same names, be aware of PATH ordering or adjust PYTHON_BIN/MY_BIN as needed.
- There are no secret/credential requirements and the scripts do not exfiltrate data. Still, inspect the scripts yourself (they are short and included) and run them as a discretionary user (not root).
- If you have limited resources or want to avoid installing heavy deps (e.g., triton/torch), consider using a smaller model, a dedicated VM, or a cloud transcription service instead.
If you want additional confidence, provide checksums for the scripts or request a signed release from the skill author before running the installer on production hosts.
Capability Analysis
Type: OpenClaw Skill
Name: openclaw-whisper-voice
Version: 1.0.0
The skill provides a legitimate utility for local audio transcription using OpenAI Whisper. The installation script `scripts/install_local_whisper.sh` sets up necessary Python dependencies (e.g., torch, openai-whisper) and configures a local environment, while `scripts/transcribe.sh` provides a safe wrapper for the CLI. Although the installer downloads `get-pip.py` from a trusted source (pypa.io) and uses aggressive flags like `--break-system-packages`, these actions are consistent with the stated goal of configuring a host-level service and do not exhibit malicious intent or hidden exfiltration logic.
Capability Assessment
Purpose & Capability
The skill claims to provide local Whisper transcription and includes installer and transcription wrapper scripts that install Python packages into ~/.local, create a ~/.local/bin/whisper launcher, and run the whisper CLI. Required binaries (whisper, ffmpeg) and the suggested configuration for tools.media.audio align with the stated use for WhatsApp/Telegram voice notes.
Instruction Scope
SKILL.md directs the operator to run the provided install script and to use the transcribe wrapper for files or inbound voice notes. The scripts only operate on specified audio files, create temporary directories for stdout-only mode, discover ffmpeg via imageio-ffmpeg, and run whisper; they do not read unrelated system files, transmit data to external endpoints, or access secrets.
Install Mechanism
Installation is manual (scripts/install_local_whisper.sh). The script uses curl to download get-pip.py from the official bootstrap.pypa.io and pip-installs many packages (including torch from the official PyTorch CPU wheel index). These sources are standard and expected, but pip-installing many packages to a user account is a significant change (disk/network) and increases the attack surface if package sources were compromised. The install writes only to the user's ~/.local and ~/.cache, not system-wide locations.
Credentials
No credentials or secret environment variables are requested. The scripts accept an optional PYTHON_BIN override (not declared as required), which is reasonable for portability. The installer and wrapper use $HOME and ~/.local as expected for a per-user install.
Persistence & Privilege
The skill does not request always:true, does not autonomously persist beyond user-run install steps, and only modifies user-local directories (~/.local, ~/.cache). It does not alter other skills' configs automatically; it suggests how to patch OpenClaw config but leaves that action to the operator.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install openclaw-whisper-voice - After installation, invoke the skill by name or use
/openclaw-whisper-voice - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Version 1.0.0 of openclaw-whisper-voice
- Initial release providing local Whisper speech-to-text transcription for audio files and inbound voice notes on the OpenClaw Gateway host.
- Includes an installation script for setting up Python dependencies, a CPU-compatible PyTorch build, and stable CLI launchers for Whisper and ffmpeg.
- Offers a shell wrapper script for reliable manual and automated transcription with support for multiple audio formats and model options.
- Provides configuration guidance for integrating with WhatsApp and Telegram inbound audio using tools.media.audio in OpenClaw.
Metadata
Frequently Asked Questions
What is openclaw-whisper-voice?
Local Whisper speech-to-text for audio files and inbound voice notes on the OpenClaw Gateway host. Use when setting up local transcription for WhatsApp, Tele... It is an AI Agent Skill for Claude Code / OpenClaw, with 193 downloads so far.
How do I install openclaw-whisper-voice?
Run "/install openclaw-whisper-voice" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is openclaw-whisper-voice free?
Yes, openclaw-whisper-voice is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does openclaw-whisper-voice support?
openclaw-whisper-voice is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created openclaw-whisper-voice?
It is built and maintained by sabyaghosh (@sabyaghosh); the current version is v1.0.0.
More Skills