Upload folder using huggingface_hub

feat(aligner): group missing-words pairs under shared border

feat(aligner): group missing-words pairs into shared border

feat: redesign audio input as Link/Upload/Record toggle;perf: use per-segment WAVs and skip audio I/O for API calls

feat: add URL input for yt-dlp audio download;perf: defer char spans to MFA injection — reduce initial HTML size

perf: use single full.wav + media fragments — eliminate 796 per-segment file writes

perf: switch segment audio to OGG and remove inline base64 — fix long recording crash;fix: cap ZeroGPU lease duration at 120s hard limit

Remove dead code and add GPU VRAM profiling

Update dev tools

Replace midpoint split with MFA-based splitting for combined/fused segments

Fix special segments (Basmala/Isti'adha) not sent to MFA or animated in mega card

Add special_type to JSON/API responses and hide animate button for transitions

Add transition segment detection for non-Quranic phrases (Amin, Takbir, Tahmeed)

Fold Word Joiner into char groups, merge extract/resegment/retranscribe chains, relax batch limits

Fix MFA char-level highlighting for decomposed Arabic forms

Update ZWSP unicode

Preserve tatweels for proper DigitalKhatt font rendering

Restructure src/ for cleaner modularity

Extract segment rendering helpers from app.py into src/ui/segments.py