Windows 10 / 11 · One-time purchase · No subscription

Cadenza

A video editor for the stage

Multi-camera editing for concerts and performance. Sync your angles by their own audio, cut, grade and export with your NVIDIA card — and open the Premiere projects you already have.

Buy Cadenza — $49.99 See what it does

Download and run. No Python, no conda, no dependencies.
Future builds included · 15 days of help getting it running.

Cadenza — Concert Project
Cadenza editing a multi-camera concert
Built for one job

Concert footage,
cut properly.

Multi-camera

Angles that line up

Drop in every camera and let Cadenza align them by their own audio — FFT cross-correlation, accurate to the frame. No clapperboard needed, no nudging clips by hand.

Leaving Adobe

Your Premiere edits open here

Export a sequence from Premiere as XML and Cadenza opens it complete: tracks, cuts, in and out points, motion, opacity, audio levels and keyframes. A 163-clip concert imports in seconds. It writes the format back out, too.

Your GPU

Fast where it matters

Compositing runs on CUDA and export uses NVENC, so a finished concert renders at roughly the speed it plays — even with 4K angles stacked on a 1080p timeline.

$49.99
One payment · yours to keep
Buy Cadenza — $49.99

Runs without an NVIDIA card, but compositing and export are much slower.
If it will not run on your machine and I cannot get it going, email within 14 days for a refund.

What is in it

Everything the cut needs.

🎵
Audio Sync
FFT cross-correlation aligns multi-camera recordings automatically. Eyeball the position first, then let it find the exact frame offset.
🎬
Multi-Track Timeline
Unlimited video and audio tracks. Drag, trim and razor cut with link-aware editing, rubber-band selection, multi-clip drag and full undo.
Keyframes
Animate position, scale, rotation, crop, opacity, volume and pan. Keyframes show on the clip and drag to retime — punch into a 4K angle over a phrase of music.
🎨
Lumetri Colour
Temperature, tint, exposure, contrast, highlights, shadows and saturation, applied in real time on the GPU.
Cross Dissolves
Add a dissolve at any cut. The outgoing clip plays past its out point and the incoming one starts early, straight from the source — as an editor expects.
🔊
Audio Mixing
Per-clip volume in dB on a Premiere-style logarithmic fader, stereo pan, mute, and waveforms normalised to 0 dBFS.
NVENC Export
Hardware H.264 and H.265 export. A full-length concert renders at about the speed it plays.
Premiere XML, Both Ways
Open sequences exported from Premiere, and write them back out for Premiere or DaVinci Resolve.
💾
Projects That Keep Everything
Clips, in and out points, effects, keyframes, transitions and track layout — every razor cut and grade comes back as you left it.
Before you buy

Questions

Do I need Python or any setup?
No. The zip contains Cadenza.exe and everything it needs. Extract it anywhere and run — nothing is installed into Windows.
Do I need an NVIDIA graphics card?
Recommended, not required. Compositing uses CUDA and export uses NVENC, so an NVIDIA card makes both far quicker. Without one it falls back to the CPU and runs, slowly.
How does it arrive?
By email, as soon as the payment clears: a download link and the password for the zip. The download is about 2.4 GB, because CUDA PyTorch travels with it.
Will it open my Premiere projects?
Export the sequence from Premiere as Final Cut Pro XML (File → Export → Final Cut Pro XML) and Cadenza opens it: tracks, cuts, in and out points, motion, opacity, audio levels and keyframes. Effects with no equivalent — colour grades, speed changes, titles — are listed rather than dropped silently.
What does support cover, and what about updates?
New builds are yours at no extra cost — the same download link serves the current one. Support is for getting Cadenza running on your machine for 15 days after purchase: it will not start, it will not see your GPU, the zip will not open. It is not training in how to edit, and not a promise to build features on request.
The source is public. Why pay?
Because building it yourself means conda, a 3 GB CUDA PyTorch install and matching a dozen pinned versions. The $49.99 buys the finished build, every future build after it, and help getting it running — not the code, which stays open either way.
For developers

Built in Python, in the open

PySide6 for the interface, PyTorch on CUDA for compositing, PyAV for decode and encode. Clone it, read it, run it from source with conda.

Source on GitHub →