A high-performance local speech recognition skill that provides lightning-fast offline transcription with word-level precision.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install whisperx
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install whisperx using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is WhisperX Speech Recognition?
The WhisperX skill is a robust speech-to-text solution designed for developers using Openclaw Skills who prioritize privacy and performance. By leveraging the WhisperX framework, this skill enables fully offline transcription that is up to 30x faster than standard OpenAI Whisper implementations. It is ideal for converting voice messages or audio files into accurate text without the need for external API keys or an active internet connection.
This skill excels at providing high-quality recognition across 90+ languages. It focuses strictly on Automated Speech Recognition (ASR), ensuring that voice data is converted to text efficiently. Because it runs locally, it eliminates latency and recurring costs associated with cloud-based transcription services, making it a cornerstone for privacy-focused AI workflows.
WhisperX Speech Recognition Use Cases
- Transcribing voice notes or audio recordings directly within an AI agent workflow.
- Implementing multilingual support with automatic language detection for global applications.
- Creating precise subtitles or searchable text logs using word-level timestamps.
- Handling sensitive audio data that must remain on-premise or local for security compliance.
How WhisperX Speech Recognition Works
- The skill receives an audio file in a supported format such as MP3, WAV, or FLAC.
- It utilizes ffmpeg to process and normalize the audio stream for the transcription engine.
- The WhisperX engine loads the specified model (e.g., base, small, or large) from the local cache.
- High-speed inference is performed locally on either the CPU or GPU to generate text outputs.
- The skill returns the transcribed text along with precise word-level time alignment for the AI agent to utilize.
WhisperX Speech Recognition Setup
To integrate this into your Openclaw Skills environment, ensure you have the necessary system dependencies and the Python package installed:
# Install ffmpeg for audio processing (macOS)
brew install ffmpeg
# Install ffmpeg for audio processing (Ubuntu/Debian)
apt-get install ffmpeg
# Install the WhisperX package
pip install whisperx
# Alternatively, use uvx for a managed environment
uvx whisperx
Note: For optimal performance on high-end hardware, ensure CUDA 12.8 is installed if using an NVIDIA GPU.
WhisperX Speech Recognition Data Schema & Taxonomy
The skill manages its data through local directories and standard output formats to ensure compatibility with various Openclaw Skills integrations:
| Component | Description |
|---|---|
| Model Cache | Automatically stored in ~/.cache/whisper/ to avoid re-downloading. |
| Supported Formats | MP3, WAV, OGG, FLAC, M4A, OPUS, AAC. |
| Output Metadata | Includes text strings, language codes, and word-level timestamps. |
| Compute Types | Supports float16 for GPU and int8 for low-memory CPU usage. |
WhisperX Speech Recognition Advanced Features
- Speaker Diarization: Distinguish between different speakers in a single audio file (requires a HuggingFace token).
- Forced Alignment: Uses phoneme-level models to ensure timestamps are perfectly synced with the audio.
- Optimized CPU Inference: Support for int8 compute types allows high-quality transcription even on limited hardware.
- Flexible Model Selection: Toggle between different model sizes (tiny to large-v3) to balance speed and accuracy.
SKILL.md
Loading
Related Openclaw Skills
Magister Skill
A specialized tool for fetching student schedules, grades, and attendance records from the Dutch Magister education portal.
ClawSec Monitor
A transparent HTTP/HTTPS proxy that inspects AI agent traffic in real time to detect threats and data exfiltration.
ClawSec Monitor
ClawSec Monitor is a transparent HTTP/HTTPS proxy designed to inspect, log, and secure all real-time traffic generated by AI agents.
Skill-Find
A discovery engine for finding and installing specialized capabilities within the Openclaw Skills ecosystem.
Skill Vetting
A specialized security framework designed to audit and validate third-party AI skills for safety, utility, and prompt injection risks.
aliyun-clawscan
A specialized security assessment tool designed to audit OpenClaw configurations and perform deep static analysis on skills to detect vulnerabilities and malicious behavior.
Featured*
