A high-performance local speech-to-text tool for Openclaw Skills featuring GPU acceleration and word-level timestamps.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install faster-whisper-local
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 faster-whisper-local 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 Faster-Whisper?
Faster-Whisper is a specialized tool designed to provide rapid, local speech-to-text capabilities within the ecosystem of Openclaw Skills. By leveraging optimized engines, it allows for high-accuracy transcription without relying on cloud services, ensuring data privacy and reducing latency. It supports a variety of model sizes, from lightweight English-only versions to the powerful distil-large-v3, making it adaptable to different hardware constraints and use cases.
This skill is particularly valuable for developers and researchers who need to integrate audio processing into their AI-driven workflows. By utilizing the Openclaw Skills framework, Faster-Whisper can be easily deployed to handle everything from simple voice memos to complex multilingual datasets, providing structured output that can be consumed by other automated tools.
Faster-Whisper Use Cases
- Transcribing long-form audio recordings locally to maintain data privacy and security.
- Generating precise word-level timestamps for video subtitling and closed captioning tasks.
- Processing multilingual audio files into structured JSON for automated data analysis.
- Implementing automated speech-to-text triggers in complex AI agent workflows using Openclaw Skills.
How Faster-Whisper Works
- The setup process configures a virtual environment and detects available NVIDIA GPU hardware for CUDA acceleration.
- Users provide audio input to the transcription script, specifying model preferences and processing flags.
- The engine applies Voice Activity Detection (VAD) if enabled to filter out silence and improve transcription accuracy.
- Audio is processed using faster-whisper models to generate transcript segments with high efficiency.
- The final output is returned in the desired format, such as plain text or metadata-rich JSON for downstream use.
Faster-Whisper Setup
Ensure you have Python 3.10 or later and ffmpeg installed on your system. To initialize this component of your Openclaw Skills, run the setup script to install dependencies and configure hardware acceleration:
./setup.sh
To transcribe an audio file, use the script provided in the repository:
./scripts/transcribe audio.mp3
Faster-Whisper Data Schema & Taxonomy
The skill organizes transcription data and model metadata to ensure compatibility with other Openclaw Skills.
| Feature | Details |
|---|---|
| Supported Models | distil-large-v3, large-v3-turbo, medium.en, small.en |
| Output Formats | Plain text (.txt) or structured JSON |
| Metadata | Includes word-level timestamps, probability scores, and duration |
| Requirements | Requires ffmpeg for audio decoding and Python 3 for execution |
Faster-Whisper Advanced Features
- Full integration with NVIDIA CUDA for significantly faster processing on compatible GPUs.
- Support for compute-type quantization (e.g., int8) to optimize memory usage and prevent OOM errors.
- Automated silence removal using Silero VAD to improve transcript quality.
- Support for distilled models that provide a 6x speedup over standard Whisper models within Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
Video STT Skill
A robust utility for extracting audio from video URLs and converting it into accurate text or subtitles using local models or cloud services.
Skill Manager
A specialized management utility designed to list, categorize, and identify functional overlaps within your library of Openclaw Skills.
pgvector
A high-performance PostgreSQL extension for storing embeddings and performing vector similarity searches in AI applications.
Skill Manager
A centralized management utility to audit, categorize, and deduplicate your Openclaw Skills ecosystem.
Leetify CS2 Analytics
Automate CS2 player statistics retrieval and tactical demo analysis using the Leetify API.
MarkItDown Skill
A powerful Microsoft utility for converting complex document formats into clean Markdown for AI consumption.
Featured*
