Beat-synced shorts, automatically.
Aura discovers trending short clips, filters them for quality, and edits them into beat-locked compilations - portrait, ready to post, every time.
Built and operated by PixelVision
Output
1080 × 1920 · 30 fps
Beat-synced
75 - 160 BPM
Pipeline
From signal to sequence in three steps
Discover
Pulls trending clips by theme - dance, landscape, cars - from public catalogs. No hashtag scraping. Trending songs and creators drive every run.
Filter
Runs pose detection, motion analysis, aspect-ratio checks, and text-overlay detection so only quality clips survive into the sequence pool.
Sequence
Detects the song's BPM and phrasing, then assembles surviving clips into a beat-synced portrait video - ready to upload.
What sets it apart
Built for high-volume, high-quality output
Theme-driven lanes
Distinct category profiles (dance, landscape, cars) keep each sequence visually coherent and on-brand for the lane it serves.
Beat-aligned cuts
Phrase-aware editing locks transitions to musical bars, with kick-drum onset detection for cleaner-feeling cuts than a fixed-interval splice.
Synced karaoke lyrics
Landscape and cars sequences pull synced lyrics from lrclib and Genius, with manual paste as the long-tail fallback. Real lyrics, never auto-transcribed guesses.
macOS-native studio
SwiftUI + AppKit + AVKit on the front, Python + ffmpeg + MediaPipe on the back. Ships as a signed .app - local, fast, and offline-friendly.
An internal tool, openly described
Aura isn't a product you can download. It's the workshop behind a slice of PixelVision's short-form output, documented here so the third-party services it integrates with - lrclib, Genius, and others - can verify a real owner is on the other end of every API call.