Audio2SRT Local Transcription Translation Subtitle Tool
Paste the following prompt into your AI chat to install this skill:
Please follow https://skillhub.cn/install/skillhub.md to install @user_af21af52/audio2srtlocal.
About this skill
Problem to Solve
When preparing local subtitles, teams often face repeated friction from cloning repositories, resolving dependencies, and downloading models. audio2srt-local targets a narrower job: generating a runnable Web GUI on macOS Apple Silicon, with a Python aiohttp transcription backend and a Vite + React frontend.
How the Skill Works
- Embedded project templates: it writes the project files directly into the target directory, avoiding a
Giteeclone or network-based source download. - Dependency setup: it separates Python and Node.js installation steps, reducing environment mismatch during local deployment.
- Model and service startup: it checks for
whisper-large-v3-turboandQwen2.5-3B-Instruct-4bit, fetches missing models from ModelScope, then starts the backend on port8765and the frontend on port3000.
Fit and Limits
The skill is for local source generation, not a universal cross-platform deployer. It expects macOS 14.0+, Python 3.10+, and Node.js 18+, and the first model pull is roughly 4GB+. If you already have a repository, need a Gitee clone, or prefer a source-download deployment flow, use the corresponding audio2srt-deploy.
Use Cases
- Turn English meeting audio into a locally accessible transcription and translation subtitle app on an Apple Silicon Mac.
- Create the Audio2SRT project and install frontend/backend dependencies on a local machine without cloning a repository.
- Process existing m4a or mp3 files and review sentence-level transcripts plus Qwen translations in the browser.
- Build an MLX Whisper transcription GUI from embedded templates and start the local services when a repo clone is unavailable.
Best For
- Video subtitle editors who want to turn local recordings into a reviewable local subtitle interface quickly.
- macOS Apple Silicon engineers deploying standalone tools who want to generate the project without cloning a repository.
- Researchers processing audio/video content locally who want an MLX Whisper transcription web GUI.
- Full-stack engineers maintaining local AI toolchains who want both the Python backend and React frontend generated.
Related Skills
Creates themed learning sites by generating flashcards, lessons, quizzes, index, and progress pages into a deployable website.
End-to-end mathematical optimization modeling with Gurobi and Python, covering data validation, code generation, solving, result analysis, and iterative revision.
Executes an Eversheet Web CDP login using externally supplied baseUrl, username, and password, then verifies the page reaches a logged-in home state.
Clarify product goals, user flows, pages, data, and permissions, then output an implementation-ready development prompt and lightweight spec.