Our platform revolutionizes digital storytelling by automating the entire YouTube content pipeline. Using Google Cloud’s powerful infrastructure, we transform raw data into engaging video content with high-quality AI narration and automated editing.
(저희 플랫폼은 전체 유튜브 콘텐츠 파이프라인을 자동화하여 디지털 스토리텔링을 혁신합니다. 구글 클라우드의 인프라를 활용하여 데이터를 고품질 AI 내레이션과 자동 편집이 포함된 영상으로 변환합니다.)
Google Cloud Run: A fully managed serverless platform used for containerized microservices to handle video rendering and data processing.
Google Artifact Registry: Secure management and storage of Docker container images for automated CI/CD pipelines.
Vertex AI (Gemini Pro): Leveraged for automated script generation and content optimization through advanced Large Language Models (LLM).
Google Cloud Text-to-Speech (TTS): High-fidelity AI voice synthesis for multi-language narration.
FFmpeg on Cloud Run: Cloud-native video processing engine for real-time subtitle overlay and media composition.
Google Cloud Storage: Scalable object storage for raw media assets and finalized video outputs.
Google Secret Manager: Secure orchestration of API keys (YouTube Data API, OpenAI, etc.) and sensitive environment variables.
Firebase / Firestore: Real-time database for managing video metadata and job scheduling status.
Our platform operates on a high-efficiency, serverless pipeline designed to automate the end-to-end video production process. The workflow is orchestrated across several Google Cloud services as follows:
Automated Data Fetching: The process begins with Cloud Run instances performing scheduled data scraping and API calls to aggregate source material, including biblical verses and real-time financial market data.
AI-Powered Scripting: The raw data is processed through Gemini API (Vertex AI). Using advanced prompt engineering, the system optimizes the input into a structured video script, ensuring the content is engaging and tailored for YouTube’s audience.
High-Fidelity Audio Generation: The finalized script is sent to the Google Cloud Text-to-Speech (TTS) API. We utilize Neural2 and Studio voices to generate natural-sounding, multi-language narrations that serve as the backbone of the video's audio track.
Serverless Video Rendering: The visual composition is handled by FFmpeg running within Cloud Run. This stage dynamically merges generated audio, relevant background imagery, and automated subtitle overlays into a high-definition (HD) video file.
Automated Publishing: Once the rendering is complete, the final output is automatically uploaded to our designated YouTube channels via the YouTube Data API, ensuring a consistent content schedule without manual intervention.