Reel Maker
AI-Powered Highlight Reel Generator
A Streamlit application that lets users upload a video and automatically generates short highlight reels using OpenAI Whisper for audio-to-text transcription to identify key moments.
Reel Maker automates the tedious part of short-form content creation: turning a full-length video into a highlight reel by transcribing audio and identifying the most relevant segments.
- Manually scrubbing through long videos to find highlight-worthy moments is slow and repetitive.
- Creators without editing experience need a low-friction way to produce short clips for social platforms.
- Audio content needs to be understood, not just visually skimmed, to find the most quotable or relevant moments.
A Streamlit front end drives a pipeline that extracts audio, transcribes it with Whisper, and uses the resulting text to select and export highlight segments.
User uploads a video through the Streamlit UI.
Audio is extracted from the video file.
Whisper transcribes the audio into timestamped text.
The pipeline identifies candidate highlight segments from the transcript.
Short reels are rendered and made available for download.
Key Architectural Decisions
- ✓Built a Streamlit interface for uploading a video and configuring reel generation.
- ✓Used OpenAI Whisper to transcribe the video's audio track into text for downstream analysis.
- ✓Generated short highlight clips from the source video based on the transcribed content, downloadable directly from the app.
- Transcription-first analysis is a strong signal for highlight detection compared to purely visual heuristics.
- Streamlit made it fast to ship a usable interface without building a separate frontend application.