Upload & transcribe
Drop any file. Whisper transcription runs at ingest — searchable, speaker-labelled, timecoded.
Text-based editing
Edit at the speed of text.
Every upload gets transcribed. The transcript sits under your timeline, word-accurate and timecoded. Select a sentence, delete it, and the video cuts itself. This is the single biggest speed upgrade in editorial work — and it's the core loop of Aperture.
The loop, in four moves.
Select the words
Highlight the sentence you don't want. That's the whole gesture. No scrubbing, no blade tool.
Cut selection
Aperture applies razor + ripple ops. Gaps close, audio re-joins, everything stays undo-safe.
Polish & publish
One click removes all silences and fillers. Captions, music duck, export. Done in minutes.
"I stopped scrubbing timelines. I read the interview, highlighted what was wrong, and the cut edited itself."
— the loop, every dayMoment search
Find footage by what was said.
Type "where do they talk about budget". Aperture searches every transcript in the library and ranks the hits — then deep-links you into the editor at that exact timecode. No proxy scrubbing. No skimming through rushes.
Click → timeline
Search results open the NLE anchored to the moment. Not a player window — the actual cut.
Search everything
Transcripts, titles, tags, and descriptions across the whole tenant library.
Span to timeline
Found it? Add that span to your sequence with one more click and keep cutting.
Feature · Text-based editing
Delete is the most precise blade ever made.
Select the sentence. Press cut. 4.2 seconds vanish, the audio re-joins clean — and nobody scrubbed a single frame to get there.
Cut a sentenceIn plain words
Your video is now a document.
You know how a text document lets you select a paragraph and press delete? That's the whole trick. When a file arrives, Aperture transcribes every word with its timecode. The transcript sits under your timeline like subtitles. The words are the video: delete the words, and the video loses those seconds. Keep the words, and it stays. No blade tool, no scrubbing, no guessing where a sentence starts.
"We tested three designs, and honestly, I don't think the first one was very good, but the rounded version tested best with clients."
selected → press cut → 4.2 seconds gone, audio re-joins clean
"Um, so — Remove all fillers and silences"
one AI Studio recipe → every "um" and dead pause removed in one op
AI Studio
Your transcript-aware assistant.
AI Studio lives in a chat drawer in the top nav — full conversation visible, every answer shown, suggestion chips included. It sees your transcript and your lanes, so its suggestions are real timeline ops you can apply — not chat noise. Admin-configurable model (needs the tenant OpenRouter key; everything else on this page works without it).
Recipe ops
"Remove all silences and fillers." "Cut everything about pricing." One click, previewed before it touches your timeline.
Highlight ideas
Long→shorts: LLM-ranked moments from your master, ready to send to a 9:16 sequence.
Auto-tagging
New assets get tags and descriptions written for them — your library organizes itself.
AI Studio, in context
A chat-style edit assistant that sees your library and your lanes — propose cuts, find b-roll candidates in the archive, draft titles.
Captions that look designed, not generated.
Auto-captions come from the same transcript as your edit — so they're already synced. Style them once; burn them in at export.
Synced by construction
The caption lane reads the transcript segments — no drift, no manual timing.
Click-to-edit
Fix a word, fix the frame. The list scrolls with the playhead.
Styled burn-in
Font, size, color, position — applied at render, so every export carries your brand.
Karaoke, word by word
One click turns the transcript into captions where each word lights up as it's spoken. Word timings from transcription survive the export — burned server-side, not a web trick.
Four presets, then yours
Pick clean, pop, boxed, or karaoke at CC generation — then tune font, size, color, and position per sequence.
- Captions on TikTok-style vertical exports, inside safe-zone guides (9:16 / 1:1)
- Slide and dip-to-black transitions between adjacent clips — same duration in preview and export
- Position and opacity keyframes, eased: capture +pos key / +opacity key at the playhead; clips animate between keys with linear, ease-in, ease-out, or smooth curves — identical in preview and in the render
- ⌘K command palette — jump to sequences, preview any asset in a real player and insert it at the playhead, run actions, keyboard-first
- Welcome screen + template gallery: Blank, Podcast 9:16, Social ad 1:1, Explainer 16:9 — lanes and canvas pre-seeded
- Brand accent auto-applied to captions and titles — your kit, your colors
- Text track for titles and lower thirds — canvas in browser, drawtext on server
Your next cut is a
paragraph away.
Open registration, no credit card. Pro adds AI generation.