- [x] [#001] **Cut / Mute sections** — select a time range and choose to cut (remove entirely) or mute (silence audio while video continues). Cut sections show as red overlays, mute sections as transparent blue overlays on the timeline over the transcript text and audio waveform. Backend: `ffmpeg -af volume=0` for mute, time-based cutting for removal.
- [x] [#002] **Silence / pause trimmer (in progress)** — detect pauses using min duration (ms) + amplitude threshold (dB), then apply detected pauses as cut ranges. Initial endpoint: `/audio/detect-silence`; UI includes filter controls and an "Apply As Cuts" action.
- [x] [#003] **Operation-level undo for batch actions** — explicit undo entry for actions like "Apply Silence Trim" so one shortcut/click reverts the whole operation, while still allowing normal fine-grained undo/redo steps.
- [x] [#004] **Grouped silence-trim zones (editable batch)** — when pauses are applied, tag them as a batch (`trim_group_id`) so the user can: (1) delete all zones from that auto-trim pass at once, and (2) still select/resize/delete individual zones independently.
- [x] [#005] **Edit silence-trim group settings after apply** — allow reopening a trim group and changing its detection settings (min pause ms, threshold dB, pre/post buffers), then reapplying updates to that group without affecting unrelated edits.
- [x] [#006] **Volume / gain control** — per-selection or global audio gain slider. Every editor has this. Descript users constantly complain it's missing. Backend: `ffmpeg -af volume=Xdb`.
- [x] [#007] **Speed adjustment (4th zone type)** — add speed zones as the fourth editable timeline/transcript zone type (after cut, mute, gain), allowing slow/fast playback per range or globally. Backend: `ffmpeg -filter:v setpts` + `atempo`. Common use case: slightly speed up boring sections.
- [x] [#008] **Cut preview** — before committing a delete, play what the audio will sound like with that section removed (pre-listen across the edit point). Pure frontend using Web Audio API — splice the AudioBuffer and play the join.
- [x] [#009] **Timeline shows output length** — deleted regions should visually collapse (or show as narrow gaps) so the user sees the *output* duration, not just the source duration.
- [x] [#010] **Transcript search (Ctrl+F)** — find words/phrases in the transcript and highlight matches. Pure frontend. Critical for long-form content. Jump between matches with Enter.
- [x] [#011] **Mark In / Out + delete (I / O keys)** — keyboard shortcuts to mark a time range on the timeline, then delete it. Faster than click-dragging words. Store the in/out points in state, `Delete` removes them.
- [ ] [#012] **Low-confidence word highlighting** — WhisperX already returns `confidence` per word. Words below a threshold (e.g. <0.6)shouldbevisuallyunderlinedortintedsotheuserknowswheretodouble-check.
- [ ] [#013] **Re-transcribe selection** — if Whisper gets a section wrong, let the user select a word range and re-run transcription on just that segment (optionally with a different model or language).
- [ ] [#014] **Optional VibeVoice-ASR-HF transcription backend (future)** — evaluate as an alternate transcription mode for long-form, speaker-attributed transcripts. Keep WhisperX as the default for word-level timestamp editing.
- [ ] [#015] **Word text correction** — allow editing the transcript text of a word without affecting its timing. Whisper gets homophones/proper nouns wrong constantly. Pure frontend state change; no backend needed.
- [ ] [#016] **Named timeline markers** — drop named marker pins on the waveform (like Resolve markers). Store as `{ id, time, label, color }` in the project. Rendered as colored triangles on the timeline canvas.
- [ ] [#017] **Chapters** — group markers into named chapter ranges. Useful for podcasts and lectures. Exportable as YouTube chapter timestamps in the description.
- [ ] [#041] **Customizable hotkeys / keymap editor (left-hand focused)** — allow users to view, remap, and reset keyboard shortcuts (transport, edit, save/export, zone tools), with a default preset optimized for left-hand reach (Q/W/E/R/A/S/D/F/Z/X/C/V + modifiers). Include conflict detection, an alternate standard preset, and one-click "restore defaults".
- [ ] [#018] **Audio normalization / loudness targeting** — single "Normalize" button that targets a LUFS level (-14 for YouTube, -16 for Spotify). Backend: `ffmpeg -af loudnorm`. Very high value for podcasters, ~2–3 hours of work.
- [ ] [#019] **Background music track** — a second audio track for background music with volume ducking. Major gap in Descript that TalkEdit could own. Backend: `ffmpeg` amix + `asendcmd` for auto-ducking.
- [ ] [#020] **Video zoom / punch-in** — scale and position the video (crop, zoom, pan). Used constantly on talking-head videos for emphasis. Backend: `ffmpeg -vf crop/scale/zoompan`.
- [ ] [#021] **Multi-clip / append** — load a second video and append it to the timeline. Even without a full multi-track timeline, "append clip" is a heavily used workflow.
- [ ] [#022] **Clip thumbnail strip** — video frame thumbnails along the timeline so users can navigate visually, not only by waveform. Backend: `ffmpeg` thumbnail extraction at regular intervals.
- [ ] [#023] **Batch silence removal** — full-file scan + remove all pauses above threshold in one click. Distinct from the manual trimmer above; this is a "fix the whole file" operation.
- [ ] [#024] **Export to transcript text / SRT only** — some users just want a clean `.txt` or `.srt` of the edited transcript without rendering video.
These aren't features to build — they're things to make more visible in the UI and README:
- **100% offline / no account required** — CapCut requires login and sends data to servers. Descript is cloud-first. TalkEdit never leaves the machine.
- **Local AI models** — Ollama support means no API costs and no data leaving the device.
- **Word-level precision** — editing by deleting words (not dragging razor cuts) is faster for talking-head content than any timeline-based editor.
- **Works on long files** — virtualized transcript + chunked waveform handles 1hr+ content that bogs down CapCut.
