Zoom crops toward center then re-scales to your export size with high-quality (lanczos) scaling β never stretched or pixelated.
Detects and follows the active speaker automatically instead of a fixed center crop or manual pan β only kicks in while zoomed in above 1.00x. Uses Kalman-filtered motion smoothing, scene-cut-aware clean jump-cuts, and (when multiple people are on screen) switches the anchor to whoever's actively talking. Needs opencv-python-headless installed, with mediapipe optional for multi-speaker detection β falls back automatically to a simpler single-face detector if those aren't available.
Detects quiet/silent gaps in this clip's audio. Nothing is cut until you click Apply, and you can Undo the cut afterward β this never happens automatically unless you turn on "Auto-remove Dead Air" in the automation settings.
Shows files from the audio/ folder, PLUS music/video already on this device (Music, Downloads, etc). Play each to check the fit, then hit Add to use it.
Records from this device's microphone. Works on PC and Android browsers alike.
"Neural" voices (edge-tts) sound noticeably more human than the "Basic quality" (Google TTS) ones β start there. No free/offline engine fully matches paid tools like ElevenLabs, but Andrew/Emma and the Hindi/India voices above are the most natural-sounding options available here.
Temporary β lives only for this session. Reuse any of these as this clip's voiceover, or delete the ones you don't need.
Drag any text directly on the preview to position it. Add as many as you need (titles, captions, callouts).
Pulls the subtitle/caption track this video already has (YouTube etc.) and drops it in as timed captions β no typing needed.
YouTube ke internal per-word timing se EXACT karaoke captions banata hai (guess nahi, real data). Source video ke liye ek baar fetch hota hai, phir sab clips isi ko reuse karte hain.
Drag the logo on the preview to position it.
Pick a tool, then drag on the preview to draw. Blur/Black/Emoji hide something. Rect/Circle/Arrow highlight something.
If you don't change anything else (speed, zoom, color, regions, text, logo, audio), Save just copies the file instantly instead of re-encoding.
Same silence-cut tool as the Speed/Zoom tab, handy right before export. Apply is manual and always Undo-able.