transcribing-media-to-minutes

Convert Thai-English meeting recordings into timestamped minutes and verbatim speaker-labeled transcripts.

Updated Apr 26, 2026
One-click install
npx skills add https://github.com/Jin9/skillify-foundation --skill transcribing-media-to-minutes-jin9
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: transcribing-media-to-minutes
Source: https://github.com/Jin9/skillify-foundation/tree/main/treasury/transcribing-media-to-minutes
Command: npx skills add https://github.com/Jin9/skillify-foundation --skill transcribing-media-to-minutes-jin9

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Turning a recorded meeting into accurate Minutes of Meeting is slow and error-prone, especially for Thai-English bilingual recordings where code-switched banking and tech terms must be preserved. This Skill transcribes the audio or video, extracts decisions and action items with timestamp evidence, and renders plain-language Thai-primary minutes without inventing details. ## Core Features & Use Cases - Faithful MOM extraction: Derives attendees, agenda, decisions, action items, open questions, and risks from the transcript only, with every decision and action bound to an [hh:mm:ss] timestamp. - Bilingual rendering: Writes Thai-primary plain-language minutes that keep English terms (KYC, API, deploy) verbatim and gloss them once, plus a verbatim timestamped speaker-labeled transcript sidecar. - Privacy-aware model routing: Classifies recording sensitivity (PDPA B.E. 2562) before upload, routes by length and complexity to fast or frontier model tiers, and falls back to an external transcript when the host cannot ingest media. - Use Case: A banking team records a Thai standup mixing English dev terms; the Skill produces a MOM file with timestamped action items and a verbatim transcript, flagging inaudible spans for review. ## Quick Start Transcribe the attached meeting recording standup.m4a and generate Thai minutes of meeting with a timestamped transcript.

Frequently Asked Questions about transcribing-media-to-minutes

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn a meeting recording into minutes of meeting?

Provide the audio or video file path and the Skill transcribes it, extracts decisions and action items with timestamps, and writes a minutes file plus a verbatim transcript beside the source media. Optional context like attendee names and agenda improves accuracy.

How to transcribe Thai-English bilingual meeting audio accurately?

The Skill preserves Thai-English code-switching verbatim in the transcript and keeps English banking and tech terms untranslated in the minutes, glossing each once in plain Thai. Long or multi-speaker recordings route to a higher-reasoning model tier for better diarization.

Can I transcribe meetings if my AI model cannot process audio files?

Yes. When the host model cannot natively ingest audio or video, the Skill switches to an external-transcript fallback: obtain a transcript from a dedicated ASR tool or supply one, then it resumes extraction and minutes generation from that transcript.

Is it safe to upload confidential bank meeting recordings to AI models?

The Skill runs a privacy gate before any upload, classifying content under PDPA B.E. 2562 and matching it to an eligible model class such as ZDR or self-hosted options. If no eligible model exists, it stops and asks rather than uploading sensitive audio.

What are the limitations of automated meeting minutes generation?

It works only on finished recordings, not live real-time transcription, and not on non-meeting media or document-only translation. It never invents owners, dates, or decisions; unknown values are marked as not stated and inaudible spans are flagged for review.