AI voiceovers with captions on every word.

Type a script, pick an ElevenLabs voice, a speech model and a pace, and Paraplex adds the voiceover to your timeline with a caption on each word, timed from the voice itself. The narration, the captions and the 3D motion live in one file, so an explainer, a YouTube Short or a product video comes out of one tool instead of three.

ElevenLabs voices built in. Bring your own key and skip Paraplex credits. Runs in your browser.

A vertical explainer made in Paraplex with an AI voiceover, its captions appearing word by word in time with the voice

Script in, narrated video out.

Four steps on one timeline. Nothing gets exported from one app and lined up by hand in another.

Script

Type it or paste it.

The narration goes into the Voice window. The same words become the voice and the captions.

Voice

ElevenLabs, built in.

Search the voice list, then pick a speech model and a pace. Your own key adds your cloned voices and skips Paraplex credits.

Captions

Timed from the voice.

Every word gets its caption from the voice's own timing, in one of four caption styles.

Export

Sound in every cut.

The voice is mixed into MP4 and WebM exports, and the captions export as SRT or WebVTT.

A native ElevenLabs voice, or your own.

Press and hold Audio track in the dock and choose Voiceover · with captions on each word…. Type or paste your script, search the list with Find a voice, and pick a speech model: Expressive for the most natural read, Turbo, Multilingual for 29 languages, or Flash when speed matters most. Set the pace, and the voiceover lands on the timeline at the playhead as an audio clip.

ElevenLabs is built into Paraplex, so the voices are there the moment you open the window, and each voiceover uses Paraplex AI credits from your account. Already pay for ElevenLabs? Add your key under Account › API keys and two things change: your cloned and saved voices appear in the same list, and voiceovers run on your ElevenLabs account instead of your Paraplex credits. The key stays in your browser: it is never saved in a project or uploaded to your Paraplex account, and Paraplex checks it with ElevenLabs when you save it. The same key also runs AI sound effects and music.

Captions timed from the voice, not guessed.

Captions come from the voiceover's own word timing, so each one appears as its word is spoken. Pick a style: Shorts sets two words at a time, big and bold; Documentary and Subtitle; or Spoken, where each word lights up as it is said.

Already have subtitles? The assistant can make captions from an SRT file, or from a script at speaking pace. When the piece is done, Menu › Export captions writes an SRT file for YouTube, LinkedIn and Vimeo, or a WebVTT file for a video on your own page.

Posting to TikTok, Reels or Shorts? See social media videos.

Voice, music and sound effects on one timeline.

The voiceover is an ordinary audio clip with its waveform on the timeline: drag it, trim it, split it, fade it in and out, or set its volume. Describe a music bed or a sound effect, such as “tense, minimal documentary underscore, 90 bpm” or “a soft whoosh”, and it arrives as another clip to mix the same way. Detect beats marks every beat of the music, so cuts and titles land on the hits.

Every track is mixed into MP4 (AAC) and WebM (Opus) exports, for pieces up to an hour, and transparent WebM keeps its sound.

Cutting to the music? More on beat sync.

Want music written to your cuts, a sound on every motion and a full mixer? See sound in Paraplex.

Automate the video creation process, not just the edit.

Most automated video editing tools speed up cutting footage you already shot. Paraplex automates the step before that: making the video. Ask the assistant for a narrated explainer and it builds the scene, adds the voiceover and captions with the same tools you would use, checks its own frames, and waits for you to Apply. Every layer and keyframe it makes stays editable.

Connect Claude Code through the Paraplex MCP server and an agent can run the same steps from your terminal: open a template, swap the words, add the voice, review the frames and export the video, each change one undo step in your open tab. One scene reframes for 16:9, 9:16, 1:1 and 4:5, so a YouTube video, its Short and a LinkedIn cut come from one build.

For YouTube automation or a weekly series, that means a script in and a narrated, captioned video out, with a human pass whenever you want one.

See how AI agents connect to Paraplex, or what the workflow looks like for marketers.

Start from a narrated template.

The AI ROI gap and Why it sells are narrated Shorts with their voiceover, word-timed captions and sound already on the timeline. Whiteboard explainer and Stat drop are ready for a voice of your own. All four are free to open: swap in your script and generate the voice.

In every voiceover.

ElevenLabs voices, built in

Your own ElevenLabs key: cloned voices, no Paraplex credits

Four speech models: Expressive, Turbo, Multilingual, Flash

29 languages with the Multilingual model

Captions timed from the voice, word by word

Shorts, Documentary, Subtitle and Spoken caption styles

SRT and WebVTT caption export

Mixed into MP4 and WebM, for pieces up to an hour

Questions about AI voiceovers

How do I add an AI voiceover to a video?

Open a project, press and hold Audio track in the dock and choose Voiceover · with captions on each word…. Type or paste your script, pick a voice, a speech model and a pace, and Paraplex adds the voiceover at the playhead with a caption on each word. To start from a finished example, open The AI ROI gap or Why it sells template and swap in your own script.

Which voices and languages can I use?

Paraplex uses ElevenLabs voices with four speech models: Expressive for the most natural read, Turbo, Multilingual for 29 languages, and Flash for the fastest result. Find a voice searches the list. With your own ElevenLabs key, your cloned and saved voices appear in the same list.

Can I use my own ElevenLabs API key?

Yes. Add it under Account › API keys and Paraplex checks it with ElevenLabs when you save it. From then on, voiceovers, sound effects and music run on your own ElevenLabs account and use no Paraplex AI credits, and your cloned and saved voices appear in the voice list. The key stays in your browser: it is never saved in a project or uploaded to your Paraplex account.

How are the captions timed?

From the voice itself. Each caption is timed from the voiceover's own word timing, so the captions match the narration word for word. Choose Shorts (two words at a time, big and bold), Documentary, Subtitle, or Spoken, where each word lights up as it is said.

Can I export subtitles for YouTube?

Yes. Menu › Export captions (SRT)… writes a subtitle file for YouTube, LinkedIn and Vimeo, and Export captions (VTT)… writes WebVTT for a video track on your own page. Each caption is one cue with its own start and end time.

Can I automate narrated videos, for example for a YouTube channel?

Yes, for motion design videos. The assistant can build a narrated, captioned piece from a description and wait for you to apply it. An AI agent such as Claude Code, connected through the Paraplex MCP server, can open a template, add the voiceover, check the frames and export the video from your terminal. One scene reframes for 16:9, 9:16, 1:1 and 4:5, so a YouTube video and its Short come from the same build.

Is Paraplex an automated video editor for camera footage?

Not mainly. Paraplex makes motion design videos: 3D scenes, type, charts and product demos, with their voice, captions and music. Video clips can sit on the timeline as layers and be trimmed, but their own sound isn't played, so for cutting talking-head or camera footage a video editor is the better tool. Many teams use both: cut the footage there, and make the narrated motion pieces in Paraplex.

Does the voiceover come through in the export?

Yes. Every audio track, the voiceover included, is mixed into MP4 (AAC) and WebM (Opus) exports, for pieces up to an hour, and transparent WebM keeps its sound. GIFs and PNG sequences have no sound.

Is AI voiceover free?

The voiceover tool is on every plan, Free included. Each voiceover uses Paraplex AI credits from your account. Add your own ElevenLabs key and voiceovers run on your ElevenLabs account instead, with no Paraplex credits used. Exports are free up to 1080p with no watermark; Pro adds 4K.

Write it. Hear it. Ship it.

Start in your browser, give your scene a voice and captions on every word, and export it for every feed.

Paraplex is a browser-based 3D motion design studio with an AI assistant that edits real layers and keyframes.

01

Describe it.

One or two sentences on this way to start.

Open Paraplex
A screenshot depicting artboards in Paraplex, the 3D motion design web app, which allow you to export multiple media formats and sizes with one click, automatically formatted for different social media platforms.

Brand and social

Video ad variations

One version per spreadsheet row: columns fill headlines, prices and photos by layer name, and Export all variants renders them in one job (Pro).

Three ways to start.

Whatever the job, a project starts from words, from a finished template, or from the files you already have. All three open in the same editor, and you can mix them in one scene.

01

Describe it.

Type what you want on the home page and click Make it. The assistant builds the scene from real layers and keyframes, and you Apply or Discard what it made.

Describe a scene

02

Open a template.

62 finished templates in 13 categories, each built for you to swap in your words, pictures, logo, colors, data, music, video or voice.

Browse templates in Paraplex

03

Bring what you have.

Drop screenshots, recordings, GLB models or Lottie files on the canvas, paste an SVG from Figma or HTML and React as a live layer, or upload a CSV for a chart.

Start with a blank scene

One scene, every screen.

Whatever the job, the same scene exports for every place it runs. Add formats as artboards beside your frame, let Responsive framing re-stack blocks for each shape, and pin your logo to its corner. Then export MP4, WebM, a GIF or a Lottie, or publish a live embed for your site.

Export to

5 shapes

16:9, 9:16, 1:1, 4:5 and 4:3, or a custom size

Safe zones for

4 feeds

TikTok, Reels, Shorts and the Instagram feed

Up to

4K at 60 fps

on Pro; free exports run to 1080p, no watermark

Landscape 16:9 · YouTube, slides and your site

Vertical 9:16 · Reels, Shorts & TikTok

Portrait 4:5 · feeds on Instagram, X and LinkedIn

Render needed · FR-03

Paraplex for

Marketers

Every channel wants its own cut. Build one scene, reframe it for 16:9, 9:16, 1:1 and 4:5 with safe zones for each feed, and keep every version on brand with a design system.

Eight types of charts against a black background: Bar chart showing vertical bars for days Mon-Fri; Line chart plotting connected points for days; Area chart filled beneath a line for days; Pie chart divided into colored segments labeled by day; Donut chart similar to pie but with a hollow center; Scatter plot with dots of varying sizes; Gantt chart with horizontal bars labeled Research, Design, Build, Launch; Timeline with labeled milestones Founded, First product, Series A, Worldwide.

Render needed · FR-04

Paraplex for

Agencies

Every client has a brand, and every brand has revisions. Keep each client's colors, fonts and motion as a design system, apply it in one step, and share a link for review.

Your brand in one step.

Keep colors, fonts, text styles, materials, scene looks and motion as a design system, and apply it to any scene at once.

Video, Lottie or a live embed.

Export MP4, WebM, GIF, PNG or Lottie, or publish an embed that plays in real time on your site and answers hover, click and scroll.

Question

Answer.

App launch (your UI) template: the camera pushes in on the input of an app interface pasted in as HTML, then eases back out to a tagline
App launch template: a question types into an app window, an answer streams in, steps slide in and a chart grows
Liquid glass nav template: a liquid glass bar bends the big headline beneath it, both inside the same Paraplex scene
One Paraplex scene on four artboards: the primary scene beside Landscape 16:9, Square 1:1 and Portrait 4:5, each laid out for its shape

Screenshot · one scene on four artboards

Coming soon: post straight to your accounts.

Not available yet. Two things are on the way for marketing teams. The first is posting finished pieces straight to your linked social accounts.

The second is a long-form YouTube workflow for 3D documentaries, educational videos and tutorials: an outline and research, a voiceover, B-roll and generated 3D scenes, edits timed to the narration with music, sound and titles, landscape and vertical renders for each platform, and titles, descriptions and hashtags written for each post. Both will be listed here when they ship.

The word 'Clarity' in large white text over a black background with abstract light patterns in warm brown on the left and cool blue on the right, showcasing the advanced 3D effects, shaders and materials shown in the Paraplex app including the liquid glass motion design template.

“Every client has a brand.”

A design system per client.

Keep each client's colors, fonts, text styles, materials, scene looks and motion as its own design system in your Library, and apply it to any scene in one step. A client's kit exports as one file to share with a teammate.

“The revisions never end.”

A link for review, a step to undo.

Copy a shareable link that publishes the piece as a web page, then refresh it after each round while the address stays the same. Notes on the canvas never export, undo has no limit, and each project keeps a History of recent saves to restore.

“The volume keeps growing.”

Templates, including your own.

Start from 70+ editable templates that each say what to swap in, or save any scene as a template of your own. Data variants make one version per spreadsheet row, and on Pro, Export all variants renders them in one job.

“Every channel wants its own size.”

One scene, every format.

Add 16:9, 9:16, 1:1, 4:5 or 4:3 as artboards beside the frame, each reframed for its shape, with safe zones for TikTok, Reels, Shorts and the Instagram feed. On Pro, one export renders every enabled format.

The client's own fonts, saved in the project

Light and dark themes, or a brand and sub-brand, in one palette

Beat detection, so cuts land on the drum hits

Notes on the canvas that never export

Transparent WebM and PNG sequences (Pro)

Live embeds the client's site fills in by layer name

AI images and 3D models from a sentence

Components that follow the master's edits

Responsive artboards in Paraplex: one scene laid out in several aspect ratios side by side

Screenshot · responsive artboards

The same Paraplex embed twice on a page, one filling the layer named Headline with Glow, the other with Ember

Screenshot · one embed, filled by name two ways

On brand, in light and dark.

Link layers to your palette and give each color a value per theme. Switch the embed's theme from your page and every linked layer follows, so the scene matches your site in light mode and in dark mode.

Bring the whole design system along, with its colors, fonts, text styles and materials, and apply it to any scene in one step.

The Design systems library in Paraplex: five design systems side by side, each with its colors, text styles, materials and motion tokens and a Use in project button, under Save from this project and Import tokens

Screenshot · your design systems

Your own code, running live inside it.

Code layers bring HTML, CSS, JavaScript or React into a scene, flat over the frame or as a card in 3D. In an embed they keep running and respond to the pointer, so a live counter or a slice of your product's UI can sit inside the 3D scene on your site.

A live HTML card with a gradient heading and a filling progress bar, standing at an angle in a Paraplex 3D scene

Screenshot · a live HTML card in a 3D scene

Camera

A camera they can move.

Let viewers Orbit, Look around like a panorama, Walk with W A S D or two fingers on a phone, or keep the camera Fixed.

States

States, not just a loop.

Idle loops, presses that play once and hold, and held poses, switched by clicks, the pointer, a delay or your page.

Keyboard

Reachable without a mouse.

Every interactive layer gets a keyboard-focusable stand-in, so visitors can reach it with Tab and press it with Enter.

Reduced motion

Calm for anyone who asks.

Visitors who ask their device for less motion see the first frame held instead of the animation.

The Publish as an embed panel in Paraplex: a live embed with its two-line snippet ready to copy, scrolling and viewer options

Screenshot · publishing a live embed

A transparent, irregularly shaped glass tube floating against a smooth, wavy background of blue, beige, black, and white hues, showcasing the 3D rendering, shaders and materials available in the Paraplex app, with customizable properties and animated keyframes, like the liquid glass effect seen in this image.
Particle logo reveal made in Paraplex: thousands of fine particles stream in from a loose cloud and pack into a logo mark on a dark ground
Logo swarm template: a swarm of fine specks packs into a logo until it reads solid, holds as a few flicker warm and cool, then bursts toward the camera
The word 'Clarity' in large white text over a black background with abstract light patterns in warm brown on the left and cool blue on the right, showcasing the advanced 3D effects, shaders and materials shown in the Paraplex app including the liquid glass motion design template.
The word 'Clarity' in large white text over a black background with abstract light patterns in warm brown on the left and cool blue on the right, showcasing the advanced 3D effects, shaders and materials shown in the Paraplex app including the liquid glass motion design template.
An app demo in the Paraplex editor: a browser frame with a cursor whose path and clicks are keyframed on the timeline

Screenshot · a cursor path over a browser frame

The AI ROI gap template: a narrated documentary Short in white clay on black, its captions following the voice
Why it sells template: a narrated vertical Short about product photos, with captions that light up each word as it is said