whisper-local-api
Secure, offline, OpenAI-compatible local Whisper ASR endpoint for OpenClaw. Features faster-whisper (large-v3-turbo), built-in privacy with no cloud telemetry, low-RAM usage footprint, and high-accuracy speech-to-text transcription. Perfect for safe and private AI agent voice commands.
Why use this skill?
Deploy a 100% offline, privacy-first Whisper ASR engine. Highly accurate, low-RAM usage, and fully compatible with OpenAI API standards.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/hantok/whisper-local-apiWhat This Skill Does
The whisper-local-api skill provides a robust, offline-first speech-to-text (ASR) engine for your OpenClaw agent. By leveraging the faster-whisper implementation with the large-v3-turbo model, it bridges the gap between high-fidelity transcription and hardware efficiency. Unlike cloud-based alternatives that transmit sensitive audio data to third-party servers, this skill ensures that all processing happens locally. It exposes an OpenAI-compatible API endpoint, making it a drop-in replacement for any application or agent configuration that expects a standard Whisper interface.
Installation
To integrate this skill into your environment, use the OpenClaw command-line interface: clawhub install openclaw/skills/skills/hantok/whisper-local-api. Once installed, navigate to the local directory (defaulting to ~/whisper-local-api) and execute the bootstrap script to prepare the environment: bash scripts/bootstrap.sh. You can launch the service using bash scripts/start.sh, and verify its operational status with bash scripts/healthcheck.sh. For advanced users, the installation path can be customized via the WHISPER_DIR environment variable before running the bootstrap process.
Use Cases
This skill is ideal for environments where data privacy is paramount, such as corporate boardrooms, legal transcription, or home automation systems handling personal voice commands. It is also excellent for high-latency environments where round-trips to the cloud for audio processing are impractical. Furthermore, developers building AI agents that require real-time voice input will benefit from the low RAM footprint (approx. 400-500MB), allowing the agent to perform transcription on edge hardware or VPS instances that lack high-end GPU resources.
Example Prompts
- "OpenClaw, transcribe the audio file located at /home/user/recordings/meeting_01.mp3 using the local whisper endpoint and summarize the key action items."
- "Configure the voice-to-text service to point to my local whisper-local-api on port 9000 and confirm the health status."
- "Start listening to the microphone input and transcribe my notes in real-time, sending the text output to my scratchpad."
Tips & Limitations
The whisper-local-api is optimized for memory safety, automatically scaling between float16 and int8 precision modes to avoid system crashes under heavy load. Note that while the model is highly accurate, it runs locally on your CPU/GPU hardware; performance is highly dependent on your host's processing power. For public-facing deployments, do not expose the local endpoint directly. Always place the service behind a secure reverse proxy using HTTPS and mandatory authentication. This skill does not perform external network calls, ensuring your data never leaves your infrastructure.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-hantok-whisper-local-api": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: network-access, file-read
Related Skills
parakeet-local-asr
Install and operate local NVIDIA Parakeet ASR for OpenClaw with an OpenAI-compatible transcription API on Ubuntu/Linux and macOS (Intel/Apple Silicon). Use when the user wants private/local speech-to-text, voice transcription setup, ASR troubleshooting, or OpenClaw voice stack configuration with Parakeet (and optional Whisper fallback).
perplexity-pro-openclaw
Connect Perplexity PRO to OpenClaw with anti-bot browser automation, bypassing Cloudflare protection via Xvfb and VNC authentication