Synthesia AI Video Editor: How to Build a Video Step by Step

Synthesia's AI video editor is a browser-based scene editor: you write a script, an AI avatar reads it, and you arrange the visuals around it in three views called Canvas, Storyboard and Timeline. A video can hold up to 150 scenes of up to 5 minutes each, so it can run up to 4 hours ^[1]. Most editing tools work on every plan, including the free Basic plan. The gates sit at the edges: MP4 download starts on Starter, clickable CTA buttons and branching start on Pro, and brand kits and live collaboration are Enterprise-only. Build the script in Storyboard, polish in Canvas and Timeline, preview for free, then generate once. That order saves credits.

What your plan unlocks before you open the editor

The editor looks the same on every plan, but what leaves it does not. On Basic, videos carry a Synthesia watermark and are shared by link or embed, with no MP4 download. As listed in November 2026, Starter is $19 a month billed monthly, or $14 a month billed yearly ($168 a year), and adds a 1080p MP4 download plus up to 3 guests who comment. Each paid self-serve account has one editor, so teammates review rather than build. The table lists the editor gates from Synthesia's pricing matrix and help center.

One more gate surprises editors. Videos with interactive elements cannot be downloaded as MP4 at all; they play only in the Synthesia player, through the share page, an embed or SCORM. So if a Pro video uses branching, plan its delivery around a link or an embed from the first scene. Credits, resets and renewals work the same whatever you build in the editor; our Synthesia pricing breakdown covers how each plan meters them.

Step 1: Write the whole script in Storyboard first

The editor has three views. Canvas shows one scene at a time with its visuals, Storyboard shows the full script in one scrollable place, and Timeline handles timing, layering and animation ^[2]. Narration stays in the script box. Timeline controls when visual elements appear, not what the avatar says. Start in Storyboard because the script drives everything else: it usually sets scene length (a scene runs as long as its longest element), animations fire on words, and captions are built from the text. Getting the words right first means fewer regenerations later.

On Starter and above, Storyboard adds quick actions on each scene: Improve writing, Shorten, Longer and Change tone, with five tones from Professional to Engaging. Pauses default to 1 second, and you can add as many as you need. Rebuilt pronunciation controls, on all plans, let you set how a word sounds by speaking it, preview it and apply it across the whole video in one click. On Enterprise, saving a pronunciation to the workspace Glossary then applies it to every voice in that language. Brand names and acronyms are where this pays off most.

Treat scenes as building blocks. A scene can be copied into another video with its visuals, audio, animations and full script, which suits a series that reuses one intro or disclaimer. Swapping a scene's layout keeps the script, though some visual elements may be lost. An element such as a logo, a disclaimer or a caption style can be pasted into several scenes at once. If you already have a deck or a document, starting from it beats retyping; our guide to Synthesia text to video covers those imports.

Step 2: Arrange each scene in Canvas and Timeline

Canvas is where you place elements: avatars, text, shapes, motion graphics, media, captions, interactivity and screen recordings. Layouts give ready-made scene designs, and non-Enterprise plans get four built-in styles named Ember, Bloom, Dusk and Crimson. Motion graphics are free and labeled beta; only graphics generated after June 2026 have editable text, and older ones must be generated again. Pick the aspect ratio early.

Synthesia offers five aspect ratios: 16:9, 9:16, 1:1, 4:5 and 5:4, set per video from the menu next to its title. Changing the ratio later rearranges nothing. You refit every on-screen element by hand, scene by scene. For a 16:9 training video that also needs a 9:16 cut, finish the landscape version, duplicate it for free, then switch the copy; generating the duplicate uses credits. Stock avatars may not appear in paid advertising without Synthesia's written consent.

If you swap in your own presenter, personal avatars come with paid plans (3 on Starter, 5 on Pro), and Synthesia creates a personal avatar or voice clone only with a live consent recording from that same person, who must be of legal age; it refuses celebrities, public figures, deceased people and anyone who has not consented, and you cannot create a photo avatar of someone else even with their permission. Our Synthesia AI avatars guide covers the avatar types and their limits.

Timeline is where pacing happens. You can reorder, rename and group layers, then stagger or chain animations so elements enter one after another. A scene runs as long as its longest element. Animation length changes by dragging, while effects keep a fixed duration. Animations and effects are triggered by a word or a pause in the script, which is why the script comes first. Point-in-time editing, which adjusts elements at an exact moment while you stay in preview, is still in beta.

Between scenes, the editor offers 15 transitions, from Fade and Wipe to Push and Direct cut. They cannot sit on scenes with interactive elements or on the final scene, and they show only when you preview the whole video. Music can come from your own library, the Soundstripe stock library or an uploaded .mp3, .wav, .ogg, .aac or .flac file. It starts at 10% volume, and Synthesia recommends 15% or less. Publishing to YouTube with Soundstripe music needs a YouTube code for each song, generated per video.

Step 3: Turn screen recordings into narrated walkthroughs

For software training, the AI Screen Recorder is a Chrome and Edge browser extension, listed on every plan. It records for up to 60 minutes, though Synthesia recommends staying under 30, and it has no resolution setting. You can also upload an MP4 made in another tool such as Camtasia or Loom, up to 500 MB and 30 minutes long. Synthesia transcribes the recording into an editable script that an avatar speaks, removes filler words, and splits recordings longer than 5 minutes into scenes. One catch: recordings that include voiceover or start from the extension get a default video template, not the brand kit or template you selected. Restyle them in Canvas.

Step 4: Set captions, brand styling and fonts

Choose captions by where the video will play. Synthesia has three types ^[3]. Closed captions are automatic, follow the script, cannot be styled and cannot be switched on by default; viewers toggle them in the Synthesia player, share page, embeds and SCORM, and they are not embedded in a downloaded MP4. Burned-in captions are switched on in the Generate panel and stay in the frame permanently. Dynamic captions, on all plans, offer 6 styles, highlight each word as it is spoken and are the only type you can style.

Brand kits, which hold up to two fonts, up to 120 colors, two logos and six avatars, are Enterprise-only. On Basic, Starter and Pro you can still set colors, fonts and a logo by hand inside a template, then reuse that template for every video. Fonts are the tricky part. The pricing page lists custom font uploads on Pro and Enterprise; Synthesia's help center describes them as Enterprise-only, so confirm before buying Pro for fonts. Both sources agree that Basic and Starter cannot upload custom fonts, including for PowerPoint imports.

Step 5: Review, preview and generate without wasting credits

Starter and Pro are single-editor accounts. Guests, 3 on Starter and 5 on Pro, can view and leave timestamped comments with replies, @mentions and reactions. Guests never edit. Live collaboration, meaning real-time co-editing of one video, is Enterprise-only. Edits autosave while you are connected, and version history lets you return to earlier autosaves. Restoring replaces the current version and cannot be undone, and the history does not log who made each specific edit, only who was editing at the time.

Preview before you generate. Previewing a scene or the whole video uses no credits, but avatars are not animated during previews ^[4]. Watching the avatar actually perform one scene before a full render needs Scene Preview, which is Enterprise-only. At generation on the self-serve plans, avatar video uses about 100 credits per finished minute (more if the avatar is on screen throughout), charged on new avatar footage only: change 3 seconds and you pay for 3 seconds. On Enterprise, video generation uses no credits. Credits reset each billing period without rollover, so batch your fixes into one regeneration.

Finish on a desktop. The editor, Avatar Builder and Brand Kits need a desktop or laptop, on Chrome 133+ or Edge 134+; mobile browsers cover dubbing, AI Playground and Syren, and there is no app. Up to 10 videos can generate at once. Every video passes content moderation at generation, and stock avatars are limited to neutral, factual content, so a rejected script means rewriting words, not moving elements. Only paid plans can appeal a rejection, and Basic cannot.

The bottom line

Synthesia's editor rewards a script-first workflow: write and tune the words in Storyboard, lay out and time scenes in Canvas and Timeline, pick captions for where the video will play, then preview for free and generate once. The editing toolkit is broad on every plan, and the walls are about output and teamwork rather than tools. Starter is the minimum for a watermark-free MP4 download; Pro adds branching for videos that stay in Synthesia's player; brand kits, shared editing and animated scene previews wait on Enterprise. Choose the plan by delivery.