ai-podcast-pipeline
por jeong-wooseok
DevOps & CloudResumen
Create Korean AI podcast packages from QuickView trend notes.
Descripción
Qué hace esta skill
Converts QuickView trend notes into Korean AI podcast packages. Handles dual-host script generation (Callie and Nick), Gemini multi-speaker TTS audio, Korean subtitle rendering, thumbnail creation, and YouTube metadata output. Supports full (15-20 min) and compressed (5-7 min) episode modes.
Covers the full production chain from raw markdown notes to upload-ready video, audio, and metadata without switching between separate tools.
Cuándo usarla
- Turning weekly AI trend notes into a Korean podcast episode
- Generating dual-voice audio with distinct male and female hosts
- Syncing and rendering Korean subtitles onto a podcast video
- Producing a 5-minute compressed edition for quick publishing
- Generating YouTube title options and description from source content
Flujo de trabajo de ejemplo
User asks: Create a full podcast episode from this week's QuickView trend file
- 1
Reads the specified QuickView markdown file from the Quartz root
- 2
Generates a full-mode conversational Korean dual-host script and saves it to archive/
- 3
Runs chunked Gemini multi-speaker TTS to produce dual-voice MP3 audio
- 4
Builds Korean SRT subtitles and renders them onto a thumbnail-backed MP4 with timing adjustment
- 5
Generates thumbnail and outputs three YouTube title options with a description
Final package: MP3 audio, subtitled MP4, thumbnail PNG, and YouTube metadata
Requisitos
- GEMINI_API_KEY environment variable
- ffmpeg installed on the system
Instalación
clawhub install jeong-wooseok/ai-podcast-pipelineEjecuta el comando de la CLI o pega el prompt en Claude Code, Cursor, Codex o el agente de programación que uses.