Synthesia: reviews and analysis

TLDR: Synthesia turns a written script into a video where an AI presenter reads it with lip sync. It’s the strongest option for internal training and corporate comms, with export into your company LMS and the certifications security will ask for. Its personal avatar requires recording yourself on webcam and submitting a consent video, so it clones you rather than inventing an anonymous character. Free plan of ten minutes a month, paid plans from $29.

What Synthesia is and how it works

You write the script, pick a presenter from the library, build the scenes with your images or slides, and the platform renders the video with lip sync. The stock actors are real people who licensed their likeness for this, and how many you get scales with the plan: nine on the free tier, over 240 on Enterprise.

The company has aimed at the same buyer for years, the learning and development team and the internal comms team of a mid-size or large business. That explains decisions that would make no sense in a tool built for creators: SCORM export to the corporate LMS, brand kits that force typography and colors on the whole team, single sign-on, and bulk personalization that produces hundreds of name-specific videos from a spreadsheet.

The second pillar is dubbing footage you already shot, across 70 languages on paid plans. The third, the one that decides who this tool is for, is the personal avatar.

What it’s like to use day to day

Creating your personal avatar takes three steps: you record yourself on webcam or phone reading a script, you record a separate consent video, and by the next business day you have your digital twin with a cloned voice. The official feature page is blunt about it, you cannot create an avatar of anyone without the explicit consent of the person involved, and that covers politicians, celebrities and historical figures too.

That requirement isn’t a checkbox on a form. In October 2024, thirty testers convened by NIST and the nonprofit Humane Intelligence made over forty attempts to get around it using celebrity faces, videos scraped from YouTube and tampered webcam feeds. The system blocked every one. For a company about to put its HR director’s face in a video, that barrier is exactly what you want. For anyone who wanted an invented presenter, it’s a closed door.

The rest of the workflow feels closer to building slides than editing video, and that’s both its strength and its ceiling: a team with no production background ships presentable material on day one, while an experienced editor hits the lack of fine control almost immediately.

Quality depends on the script as much as on the avatar, and the bar moves with the shot. In mid-shot with informational content, the top-tier presenters pass without anyone stopping to study them. In close-up with strong emotion, they don’t: a hands-on test by MIT Technology Review of the latest generation described pink-toned palms, stiff hair and eyes that barely blink. None of that matters in a workplace safety module. It costs you in the CEO’s message after a bad quarter.

Pricing and plans

The model is freemium: a permanently free Basic plan, and paid plans from $29 a month, dropping to $18 if you pay for the year up front. The next step up, Creator, runs $89 monthly or $64 on annual billing. The free tier shows you how it works and little else: ten minutes a month, a watermark, no file download and no personal avatar, which arrives with Starter.

What deserves a hard look before you commit is the minute counter, the most repeated complaint among people paying for it. Starter gives ten minutes a month and Creator thirty. An eight-minute training module eats almost the entire monthly allowance on the entry plan, and the takes you end up discarding count all the same, because usage is measured when you render rather than when you publish. If you iterate the script three times, budget for triple.

The other jump that catches buyers out is team size. Starter and Creator include a single editor, and everyone else joins as a guest who can comment but not touch the video. Producing with two people means moving to Enterprise with custom pricing.

Who it’s for (and who it isn’t)

Synthesia is for companies: internal training, compliance, standard operating procedures, HR communications. If you need the same content in several languages, refreshed every quarter without booking anyone back into a studio, and published in the corporate LMS with tracking on who watched it, nothing in this category fits better.

It isn’t for anyone who wants a synthetic character. If the plan is a channel fronted by a presenter who doesn’t exist, or publishing without showing your face, Synthesia shuts that door by design, because its personal avatar is your face with your consent on record. That ground belongs to HeyGen, whose generated avatars don’t start from a real person, and to Hedra, which animates a character from a single image. It’s a poor fit for high-volume short video too: the minutes run out fast and the cost per minute on entry plans sits among the highest in the category.

One warning belongs on the record. Stock actors license their likeness under broad contracts, and in 2022 several found their faces in government propaganda. The company acknowledged the moderation gap and tightened its policies. If you build your brand’s communications on a library face, keep in mind that other companies are using that same face.

Alternatives to Synthesia

HeyGen is the direct rival and wins on flexibility for creators and on dubbing that keeps your own voice. Creatify plays a different game, video ads generated from a product page. And Hedra gives you animated characters from an image. They’re all in Synthesia alternatives, and the full category sits in our guide to the best AI video tools.

Frequently asked questions

Can I create an avatar of someone else?

Not without their explicit consent. The process requires a consent video from that same person, and the identity has to match the recorded material. Attempts to work around it with celebrities or scraped video get blocked.

Is the free plan enough to work with?

Enough to try the tool, not to publish. It gives ten minutes a month with a watermark, no file download and no personal avatar.

Can you tell the presenter is AI?

It depends on the shot. In mid-shot with an informational script, most viewers don’t pause. In close-up, with silences and emotion, the fixed gaze and repeated gestures give it away.

How many video minutes do I need per month?

Count what you’ll render rather than what you’ll publish, because every discarded take burns allowance. One module a month with two rounds of revisions already runs short on the entry plan.

Does it dub footage I already shot?

Yes, across 70 languages on paid plans, preserving the original voice and accent. Lip sync is optional and costs double the credits.