Transcribe Korean meeting audio into structured JSON with optional speaker diarization.
npx skills add https://github.com/tellang/sonote --skill sonote
Please help me install this Agent Skill. Skill: sonote Source: https://github.com/tellang/sonote/tree/main Command: npx skills add https://github.com/tellang/sonote --skill sonote
This skill helps you turn real-time Korean speech from meetings or YouTube live streams into accurate, agent-ready transcripts without manual transcription effort.
Use case: You run an agent-driven workflow for a professional meeting—capture speech, get speaker-separated transcripts in JSON, then automatically produce a structured summary and action items for follow-up.
Run the skill in meeting mode with diarization and JSON output by starting a local server for real-time subtitles at 8000 using: sonote meeting --json.
High-intent search queries and answers about installing and using this skill.
Real-time Korean meeting transcription with speaker diarization is produced by capturing microphone input and using an HF_TOKEN-powered diarization process to separate speakers. You can run the meeting mode with JSON output enabled to generate structured, agent-ready transcripts for downstream automation.
Yes, you can transcribe YouTube live streams into structured JSON by applying the tool to continuous YouTube audio workflows. It streams low-latency transcription output that is SSE and WebSocket-ready, making it suitable for downstream automation agents.
You need ffmpeg installed as a required preflight tool to process audio for Korean STT, while an optional HF_TOKEN is required specifically for enabling speaker diarization. Failures in these preflight checks map directly to typed exit codes.
LLM post-processing refines meeting transcripts by sending the initial Korean STT output to supported LLM backends like Codex, Gemini, or Ollama. This automatically generates refined text, structured summaries, and to-do items for meeting intelligence.
The best way to capture offline audio and video for Korean transcription is using the scan-to-download workflow. It processes local or offline media files into machine-readable text with optional speaker diarization, returning structured JSON results.
Faster-whisper supports low-latency streaming output for Korean speech-to-text workflows. When integrated with the FastAPI and uvicorn server setup, it produces streaming outputs that are SSE and WebSocket-ready for real-time agent applications.
harshav167Transcribe live microphone or pre-recorded audio into text using Apple's Speech framework.
beshkenadzeConvert meeting transcripts into structured notes with decisions and action items.
RuifengFuExtracts text from Bilibili videos and subtitles using yt-dlp, ffmpeg, and faster-whisper ASR.
swyxioConvert video inputs into timestamped markdown timelines with screenshots and transcripts.
