Description and Subtitle Editor Quick Start Guide

What the editor does

The editor creates, times, reviews and narrates three event types: audio description for essential visual information, subtitles for speech and relevant sound, and review notes for issues that must be checked by the team. Events can be draft, in review or approved, and may include a speaker, narration notes, tone and pace.

Quick start

Workspace and projects

The workspace contains project actions, source video and expert timeline, the chronological event list, TTS/mix settings, human narration, ContentStudio and quality-control panels. The visual timeline shows waveform, scene changes, markers and event blocks; keyboard users reach the same positions with playback and navigation commands.

Save: Alt+Shift+S asks for a name the first time and updates the same project afterward. Save As creates a copy. Autosave runs periodically while a saved project is open.

Timeline and markers

1. Press Alt+T to focus the timeline.
2. Play with Space and navigate with arrows or Page Up/Page Down.
3. Press M at the start and again at the end.
4. Shift+Space previews the marked range. Press D, S or A to create the event.

Alt+Left/Right moves to the previous/next marker. Alt+Up/Down changes the fine-seek step. Scene analysis can flag events that cross a cut.

Adding and editing events

Ctrl/Cmd+Enter opens the general event form. D, S and A open it with the relevant type selected and focus the text field. Enter adds a line; Ctrl/Cmd+Enter saves and returns to playback.

Each event can contain time codes, text, workflow status, speaker, narration note, tone, pace, voice, playback rate and mix levels. The editor displays duration, word count and estimated reading time; automatic fitting may increase speech rate up to the configured safe limit.

Event list

Alt+L focuses the list. Arrow keys select and preview events, including overlapping narration. Right Arrow opens the action menu; Left Arrow or Escape closes it. Shift+arrows extends selection, Ctrl/Cmd+Space toggles an item, and Ctrl/Cmd+A selects all.

Events can be moved by 1 ms, 10 ms, 100 ms or one second without moving the source video. Editing and saving an already narrated item regenerates only that item, even if the text was not changed.

Import and export

Import SRT or VTT as subtitles or descriptions, then edit them like native events. Export all or selected items and filter by type.

• SRT/VTT: time codes and visible text for subtitle or studio tools.
• Excel: event type, status, speaker, text, narration notes, tone and pace.

Ctrl/Cmd+F opens search; Ctrl/Cmd+H opens find and replace.

Narration and mixing

Default settings support system voices, OpenAI, ElevenLabs and human narration. Cloud services require their API keys in EVD AI settings. Set narration rate and level, original-video level (default 90%), and the maximum automatic fitting rate. Every event can override these defaults.

Preview mixes narration with the original video using the event's settings. Create described video or MP3 places TTS or approved human takes at their time codes. Video supports visual review; MP3 is useful for sharing the completed audio mix. Events without generated or matched audio remain silent in the output.

Human narration matching

Use a clean MP3/WAV recording containing only the narrator. EVD can find candidates with local Whisper or let you match every take manually. Both methods may be combined in one project.

Automatic candidate matching

1. Select Tiny (fast), Base (recommended) or Small (more accurate and slower).
2. Choose Select and match clean human narration. On first use the model is downloaded and cached; later analyses use the local cache.
3. Use Up/Down in the candidate list. Space plays the take; Ctrl/Cmd+Space plays it mixed with the film.
4. Enter or Alt+Enter applies the take. Right Arrow opens listen, mix, edit and apply commands. High-confidence matches may be applied together with Use recommended takes.

Errors, coughs, spoken production notes and alternate readings remain available for editorial review rather than being forced into the project.

Trim a candidate

Press Ctrl/Cmd+Enter on a candidate. K plays/pauses; J/L seeks one second and Shift+J/L seeks 100 ms. Set start and end with M, preview the selection with Shift/Ctrl+Space, and save with Enter.

Manual matching

1. Open Match manually. EVD selects the first event without narration and announces its text.
2. Alt+T focuses the narration timeline. K/Space plays; J/L seeks one second and Shift+J/L seeks 100 ms.
3. Ctrl/Cmd+Left/Right or Ctrl/Cmd+J/L jumps between speech starts and plays a short sample when paused.
4. Press M at the start and end of the reading. Shift/Ctrl+Space previews it; Enter assigns it and advances to the next event.
5. Shift+M clears only active markers. D announces the current text. Alt+L returns to the match list; Delete removes an incorrect assignment, while Alt+Enter approves an available automatic candidate.

Use Shift+arrows or Ctrl/Cmd+A for multiple selection. F/A changes narration speed, Alt+Shift+L/J narration volume, and Alt+Shift+Up/Down original-film volume. Changes apply to all selected events and are previewed immediately.

ContentStudio automatic description

Save the API key once, check account/credit, choose language, template, detail, processing mode, confidence and creativity, then add production-specific instructions. Upload the video and monitor progress. When ready, import results as drafts and review every text and time code. A previous remote job can be recovered with Find last matching project.

Quality control

Quality control reports invalid time codes, empty text, unresolved notes, unfinished workflow states, missing or outdated narration, text that does not fit, suspicious gaps, overlapping events/TTS, scene-crossing events and low-confidence AI drafts. Select a result and press Enter to reach its event; reports can be saved as TXT or Excel.

Keyboard shortcuts

Playback commands work only when the relevant timeline is focused, preventing accidental playback while typing or activating controls. Additional commands can be assigned in the Keyboard Shortcut Manager.

Troubleshooting and good practice

Narration does not fit

Shorten the text first, then check automatic fitting and its safe maximum. Extend or split the event rather than using unintelligible speed.

Two descriptions overlap

Preview includes neighboring narration. Move the event in fine steps and run quality control again.

Human matching is weak

Use a clean narrator-only file, try Base or Small, and listen to candidates in the film mix before accepting them. Use manual matching for uncertain passages.

Recommended final sequence

1. Finish drafts and workflow states.
2. Run scene analysis.
3. Preview every narration in the film mix.
4. Resolve quality-control blockers and review notes.
5. Save the project, archive an Excel report, and create the final video or MP3.

Back to top