fix: video merger transition & smooth mode - clamp transition duration, prevent video start cut

fix: add yuv420p + faststart for mobile video compatibility

feat: add Veo watermark removal (delogo) option to video merger

feat: add HF Space TikTok proxy relay to bypass Vercel Cloudflare blocks

feat: add Video+Music combine module, GPT Image for Product Photoshoot, fix flow config

feat: add /api/combine-audio endpoint for video+music combining

feat: add intro text overlay (first 3 seconds) to video merger

feat: add Remove BGM toggle - isolates vocals using FFmpeg highpass+lowpass+afftdn filters

feat: uppercase subtitle mode reduces font size 15% and forces bold

feat: add UPPERCASE subtitle toggle option

fix: add README.md for HF Spaces configuration

fix: karaoke word data lost in build_whisper_subtitles + chatgpt empty response guard + parser markdown bold

fix: karaoke spotlight - only 1 word colored at a time like CapCut, not all words

feat: karaoke word-by-word highlight effect using ASS kf tags synced with Whisper timing

feat: Whisper-powered word-level subtitle like CapCut, auto-transcribe from video audio

fix: subtitle pure dialog text, bigger font 55-65, chest-level position MarginV=620

feat: CapCut-style subtitle burning with 7 template styles via HF Space