preserve
Preserve the original
Keep the source image and audio as the authority before any trim or transcript pass.
Raw human footage, made postable.
Transcript-led clipping for real recordings, original image, original audio, and human texture that should not be overproduced.
00:14 Keep the weird real-life setup.
00:39 Cut the silence, keep the laugh.
01:28 Export the part that feels human.
preserve
Keep the source image and audio as the authority before any trim or transcript pass.
transcribe
Deepgram or another provider creates searchable captions against the complete raw video, not just the export.
trim
Laravel queues call media workers for auto-editor, ffmpeg, or whisper-assisted cuts while keeping customer state in Laravel.
export
Return vertical originals, transcript sidecars, captions, and light sound accents without fake avatars or synthetic polish.
Simple Stripe spine
Credits reset every month. Unused credits do not roll over.
One credit equals one raw processing minute. Laravel owns the ledger, uploads, transcript records, and export state; media workers only process lanes.
$19/mo
60
raw minutes / month
One person turning raw phone recordings into weekly clips.
$49/mo
240
raw minutes / month
Creators with repeat recording days and multiple social exports.
$149/mo
900
raw minutes / month
Small teams processing interviews, founder clips, and recurring life footage.
Cut the losers
No fake avatar packs. No clone voice. No plastic ad energy. Human keeps the awkward pause when it makes the clip better and removes it when it wastes the viewer.
The product wins by owning the transcript, the raw file, the cut list, and the export ledger. Stripe plans come after that spine, not before it.
Product spine
reverb
Live transcription progress, export-ready events, and reviewer presence move through private and presence channels once Reverb is installed.
precognition
Validate file type, duration, orientation, consent, and credit estimate before the upload becomes a job or burns a processing minute.
scout-database
Scout waits for recording and transcript models, then indexes the full transcript instead of only the final clips.