Which audio and video file formats are supported?

FunASR Transcribe supports a wide variety of common media formats, including MP4, MOV, MP3, WAV, M4A, and FLAC.

Will the transcription service drain my system resources?

The service is designed to be efficient. It only loads models when a request is received and automatically shuts down after 10 minutes of inactivity to free up system memory.

Can it distinguish between different people speaking?

Yes, by enabling the diarization flag (--diarize), the skill can identify and label different speakers in the output Markdown file.

Does this skill require an internet connection for transcription?

The core transcription is performed locally using FunASR models on your machine. However, the optional AI summarization feature requires an internet connection to communicate with Claude's API.

How do I install the necessary models?

You can use the provided setup script (python scripts/setup.py) which automatically checks your environment, installs dependencies like PyTorch, and downloads the required Paraformer models.

FunASR Transcribe

Name: FunASR Transcribe
Author: cat-xierluo

bycat-xierluo

•

Productivity & Workflow

Transcribes audio and video files into structured Markdown documents with timestamps and speaker identification using local FunASR models.

FunASR Transcribe is a specialized Claude Code skill designed to streamline the conversion of multimedia files into readable, searchable Markdown text. By leveraging local FunASR services, it ensures data privacy while providing advanced features like automatic speaker diarization and AI-powered summarization. It is particularly useful for legal professionals, researchers, and developers who need to generate accurate meeting minutes, video subtitles, or podcast transcripts directly within their local terminal environment without relying on third-party cloud transcription services.

Key Features

01Local model execution for enhanced data privacy and security

02Automatic timestamp generation and multi-speaker diarization

0356 GitHub stars

04Supports multiple formats including MP4, MOV, MP3, WAV, M4A, and FLAC

05Smart service management with automatic startup and idle-based shutdown

06Integrated AI summarization for key insights, speaker viewpoints, and keywords

Use Cases

01Transcribing video content or podcasts for documentation and content repurposing

02Converting legal hearings or interviews into searchable, structured Markdown transcripts

03Generating detailed meeting minutes with speaker attribution and automated AI summaries

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add cat-xierluo/legal-skills funasr-transcribe

For use in Claude.ai and ChatGPT

Download Skill