fix(denoise): rewrite step 9 — extract audio only then remux, avoids ffmpeg 6.1 filter_complex video mapping bug

fix(step9): split denoise+silenceremove into 2 passes — anlmdn chained filter crashes ffmpeg 6.1.1

fix(v3): burn captions in ffmpeg, denoise voice, jump-cut pauses, title card before footage, fix headline overlap

fix: merge duplicate env: blocks in Upload step — was causing startup_failure

feat: Bright Studio style — Space Grotesk 100px top, captions, trim, music, navy border rule

perf: 720x1280@24fps + chrome-headless-shell + each clip on own track + system fonts (target render 5-8min)

fix: HyperFrames needs project directory with index.html, not direct file path

feat: add HyperFrames testimonial pipeline — composition, workflow, patcher, README

fix: pass HF_TOKEN as Space secret in deploy workflow so uploader.py can authenticate

ci: update HF Space name to hyperframes-video-studio

ci: add HuggingFace auto-deploy workflow + git configs