KwaFlux feature
Edit subtitles on the timeline — locally, with a live preview
Subtitle Edit is the shipped desktop workbench: add cues on any range, import SRT/ASS, style and position them, then burn in or export a soft subtitle. Translation, ASR, and subtitle enhance are not released — this page does not claim them.
- Manual add or import SRT / ASS
- Live preview of style and position
- Burn-in or MKV soft-sub export — no cloud caption SaaS
Cue editor
Place captions on the real picture, then export burn-in or soft-sub
The current workbench is an editor, not a translator. Preview timing on a short slice before you export.

What is shipped today
Cue editing
- Add or import SRT/ASS, adjust in/out, style, and on-picture position with a live preview.
Burn-in or soft export
- Burn captions into the picture or export MKV soft-subs. Pick the path your client actually asked for.
Not shipped: translate / ASR
- Subtitle translate, speech-to-subtitle, and subtitle enhance remain disabled in the desktop catalog. Do not plan a localization job on those yet.
An editor, not a caption SaaS
If you already have a script or an SRT, Subtitle Edit keeps the language track on the workstation. That is the honest job today.
If you need machine translation or speech-to-text, that is not in the current release. Use a specialist and bring the SRT back here to burn or mux.
Preview timing on the worst line
Fast speech and leftover burned-in titles are the hard cases. Check a 1s/3s/5s preview on the densest line before a full export.
How to edit subtitles in KwaFlux
01
Import the picture
Open Subtitle Edit and import the video you will caption.
02
Add or import cues
Type cues or import SRT/ASS. Style and position on the live preview.
03
Export burn-in or soft-sub
Choose burn-in or MKV soft-sub. Translation is not part of this export.

System requirements
- Operating system
- Windows 10 (64-bit, 1809+) or Windows 11
- macOS build coming — planned minimum macOS 13 or newer (Apple Silicon and Intel).
- GPU
- DirectX 12 compatible discrete GPU (NVIDIA / AMD / Intel Arc)
- NVIDIA RTX recommended — TensorRT acceleration is available on RTX. macOS GPU requirements will ship with the Mac build.
- Memory
- 16 GB RAM or above
- Keeps previews and export staging responsive under load.
- Storage
- Fast SSD workspace
- Better for caching, temp artifacts, and larger preview sessions.
Windows 10 (64-bit, 1809+) or Windows 11 today; macOS coming — planned minimum macOS 13+.
Frequently Asked Questions
Can KwaFlux translate my captions?
- Not in the current release. Subtitle translate and speech-to-subtitle are disabled. Edit and export SRT/ASS or burn-in only.
Hard-burned source titles?
- Editing a new track does not erase burned-in titles. Treat any erase path as a separate, best-effort job and preview it.
Is my footage uploaded to the cloud?
- No. The AI processing runs on your own GPU and your media files are not uploaded — there is no queue and no per-minute cloud fee. The network is used only for sign-in and license checks (needed to export on a paid plan) and for updates.
Related workflows
Video Converter that lives inside your enhancement pipeline
MP4, MKV, and MOV containers with H.264, H.265, or AV1 encoding
Video Converter that lives inside your enhancement pipelineAI Video Quality Repair — without uploading a frame
Target compression damage, grain, and general softness
AI Video Quality Repair — without uploading a frameAI Audio Denoise — because half of video quality is sound
Remove background noise from video or standalone audio files
AI Audio Denoise — because half of video quality is soundGuides & comparisons
AI Audio Denoise
Remove background noise from video or standalone audio files
Learn how it worksShort-drama localization
Localize short drama on one workstation
Learn how it worksShort-drama localization on one GPU: the middle-processing steps that never need a cloud pipeline
Volcano Engine's AI MediaKit pitch is right about the bottleneck — versioning, not shooting — and wrong that it needs a cloud. What one workstation covers, and what it does not.
Learn how it worksHow to remove background noise from video — locally
Viewers forgive soft pixels before they forgive hissy dialogue. Denoise, cut room echo, and separate vocals in the same local pipeline as the picture — no extract-then-reimport.
Learn how it works