This is a submission for the GitHub Finish-Up-A-Thon Challenge

What I Built

Ultra Creative Suite is a professional video and audio editor for Windows, built from the ground up for blind and low-vision users. Every feature works equally for sighted and non-sighted users — no compromises, no stripped-down accessible mode.

I am blind. I use JAWS for Windows daily. I built this because no professional video editor on the market — Adobe Premiere, DaVinci Resolve, Final Cut — is actually usable with a screen reader.

The project includes an AI video generator that takes a single audio file and produces a complete 4K video: lyric analysis, automatic stock footage download, mood-based color grading, ambient sound mixing, and GPU-accelerated rendering.

Demo

GitHub: https://github.com/demirajvazi10-max/Ultra-Creative-suite

The demo video in the repository was created entirely by me — blind, using JAWS only. AI analyzed the lyrics, selected stock footage, applied color grading, mixed ambient sounds, and rendered to 4K. Zero sighted assistance.

The Comeback Story

The project existed and worked, but had several critical issues that made real-world use unreliable:

Before:

• Audio fade timing was broken — afade filter used absolute timestamps instead of stream-relative timing, causing ambient sounds to fade incorrectly
• totalDuration was not adjusted for AudioStartSeconds, so videos were consistently shorter than expected (2:57 audio rendered as 2:49)
• SmartOutroPool had no cap on clip repetition, causing the same outro clip to repeat endlessly
• Temp files were written directly to D:\ drive instead of system temp folder, cluttering the disk
• IskraSync used incorrect timing references, causing beat-locked edits to drift

After (with GitHub Copilot):

• afade now uses st=0 relative to stream start — ambient sounds fade correctly
• totalDuration is now corrected immediately after beat detection: rawDuration - AudioStartSeconds
• SmartOutroPool has repetition cap, recycle pool, and ultra-safe fallback queries
• All temp audio batches now go to %TEMP%\UVE_MixAudio_* with automatic cleanup after render
• IskraSync rebuilt to use timelineSpan as primary measure with proper beat-lock offset

GitHub Copilot helped me understand the scope of changes across three files simultaneously and generated a precise commit description that captured every fix — something that would have taken me much longer to write manually while navigating with a screen reader.

My Experience with GitHub Copilot

As a blind developer, navigating code with a screen reader is significantly slower than visual inspection. GitHub Copilot changed my workflow in a concrete way.

When I modified three files — AIVideoCreator.xaml.cs, RenderEngine.cs, and SkiaAnimationEngine.cs — Copilot generated this commit description automatically:

"Multiple fixes across AIVideoCreator, RenderEngine and SkiaAnimationEngine: correct audio fade timing (afade) relative to asetpts/stream start; adjust totalDuration by subtracting AudioStartSeconds so -ss trimming doesn't shorten output unexpectedly; implement SmartOutroPool with max clip duration, recycle pool, ultra-safe fallback queries and fade-out tagging..."

That level of detail would have required me to re-navigate every changed file manually. Copilot gave me that summary instantly, which I could then use directly as a commit message.

For accessibility, this is not a convenience — it is a genuine productivity multiplier.