A fully offline project generator and deployer that builds a local MLX Whisper audio transcription and translation Web GUI tailored for Apple Silicon.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install localaudio2srt
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 localaudio2srt 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 audio2srt-localgen?
The audio2srt-localgen skill provides an automated, secure way to bootstrap the Audio2SRT project directly on your macOS machine. Unlike standard deployment tools that clone repositories from external sources like Gitee, this skill leverages embedded local templates to generate the entire source codebase without requiring external network access for code generation. This makes it an ideal choice for secure, private environments utilizing Openclaw Skills.
Once generated, the system automates the installation of Python and Node.js environments, downloads highly-optimized Apple Silicon models (MLX Whisper and Qwen translation) from ModelScope, and starts up a reactive, modern Web GUI. It provides a seamless transition from code generation to a fully functional transcription and translation server right on your Apple Silicon Mac.
audio2srt-localgen Use Cases
- Local speech-to-text conversion and transcription tasks on macOS using optimized MLX Whisper models.
- Deploying an offline Web GUI for audio translation and subtitle generation (SRT format) with Zero-Git requirements.
- High-security environments where cloning source code repositories from external platforms like Gitee is restricted, making Openclaw Skills local generation critical.
- Developers needing a quick, templated setup of a combined Python (aiohttp) and React (Vite) developer environment tailored for MLX-based workloads on Apple Silicon.
How audio2srt-localgen Works
- Directory Selection: The agent prompts the user for a destination directory, defaulting to
~/audio2srt. It verifies permissions and handles directory cleaning if requested. - Local Code Assembly: Using references from Openclaw Skills, the agent dynamically writes all required configuration, backend, and frontend files sequentially to maintain dependency integrity.
- Environment Setup: It installs Python dependencies specified in
server/requirements.txt(creating a virtual environment if needed) and runsnpm installfor frontend React packages. - Model Provisioning: The execution engine runs
start.sh, checking for local MLX models and automatically utilizing ModelScope CLI to download missing Whisper and Qwen weights. - Launch & Listen: It starts the Python backend on port 8765 and the Vite React development server on port 3000, presenting a ready-to-use local web client.
audio2srt-localgen Setup
To install and execute the local code generation and deployment process, trigger the skill via your Openclaw Skills workspace and use the following steps:
Prerequisites
- macOS 14.0+ on Apple Silicon (M1/M2/M3/M4)
- Python 3.10+ and Node.js 18+ installed on your host system
Step-by-Step CLI Execution
- Run the local generation trigger inside your agent terminal. Once the project structure has been scaffolded at your target directory (e.g.,
~/audio2srt), navigate to it:
cd ~/audio2srt
- Setup the Python environment and install backend dependencies:
python3 -m venv venv
source venv/bin/activate
pip install -r server/requirements.txt
- Install the Node.js frontend dependencies:
npm install
- Execute the startup script to download the MLX weights and launch both servers:
./start.sh
Once the servers boot up, open your browser and navigate to http://localhost:3000 to start transcribing audio files locally.
audio2srt-localgen Data Schema & Taxonomy
The audio2srt-localgen skill structures the generated project cleanly and logically across frontend, backend, and model directories:
Project Tree Structure
| Directory / File | Description |
|---|---|
server/ |
Python backend module |
server/transcribe_server.py |
Python aiohttp server implementing MLX Whisper & mlx-lm |
server/requirements.txt |
Python backend dependencies |
src/ |
Frontend React + TypeScript client applications |
src/components/ |
Custom components (DropZone, ResultPanel, FileCard, SrtTranslatePage) |
src/store/queueStore.ts |
Zustand queue and state management configuration |
models/ |
Storage location for heavyweight MLX model directories (.gitignore protected) |
start.sh |
Orchestration script managing ModelScope download & service runtimes |
Model Mapping
- Transcription Engine:
mlx-community/whisper-large-v3-turbo-4bitmapped locally tomodels/whisper-large-v3-turbo(approx. 2.5GB) - Translation Engine:
mlx-community/Qwen2.5-3B-Instruct-4bitmapped locally tomodels/Qwen2.5-3B-Instruct-4bit(approx. 2.0GB)
audio2srt-localgen Advanced Features
- Zero-Network Code Scaffolding: Generates the entire React, Vite, and Python application from embedded templates without invoking Git commands or remote API clone endpoints.
- ModelScope Auto-Bootstrap: The start script automatically provisions the 4GB+ MLX models from ModelScope's mirror, optimizing download speeds within region-restricted developer networks.
- macOS Native Audio Conversion: Integrates natively with Xcode CLI's
afconvertsystem utility, enabling local automatic conversion of complex container audio types like M4A and MP3 into PCM WAV buffers before inference. - Independent Service Port Management: Dual-service orchestration running frontend routing on Port 3000 and backend API routing on Port 8765, allowing clean scaling, custom proxies, and decoupled frontend development.
- Flexible Troubleshooting Hooks: Includes pre-packaged commands within Openclaw Skills integrations to easily kill zombie processes, clear corrupted NPM lock files, and force clean reinstalls of the environment.
SKILL.md
Loading
Related Openclaw Skills
Brief Master
An elite execution brief generator designed to eliminate wasted tokens and ensure precise downstream agent execution.
Strict Father AI Astrology Diagnosis
An uncompromising self-improvement skill that uses the Ziwei Doushu astrological system to expose blind spots and shatter self-deception.
OpenSCAD Automation
A powerful skill enabling AI coding agents to write, render, and export programmatic 3D CAD designs directly through the OpenSCAD command line.
Visual DNA
Extracts visual identity from screenshots, images, or URLs into a reusable Design DNA JSON specification to generate faithful UI implementations.
AI Code Review
An AI-powered multi-dimensional code review tool that scans code changes and provides structured feedback, risk assessments, and code fixes.
OPC Playbook (One-Person Company)
The OPC Playbook is a dual-mode strategic framework designed to help solo founders rigorously evaluate startup viability and systematically execute an AI-native business model.
Featured*
