AI Spokesperson Video: Turn a Script Into an On-Camera Presenter

An AI spokesperson is a presenter you build once in VdoBloom and reuse. You define a persona, generate their face, save it, then feed in a script and get a talking-to-camera video. The same saved face comes back for every future script, so your brand keeps one consistent presenter without hiring or filming anyone. Generation takes about 2 to 4 minutes.

Updated October 2026.

What you are actually building

Most AI video tools ask you to describe a whole scene every time. The spokesperson tool works the other way around. You spend a few minutes up front defining a presenter, and from then on you only write the words. The face, the style and the framing stay put. That is what makes it read as a brand presenter rather than a one-off clip: a viewer who sees the same person in video three as in video one trusts the channel more.

The persona image is generated with the platform’s image models, and the finished spokesperson is saved in your browser under a brand-spokesperson key. Every future video you make in this tool pulls that same face, which is the whole point. If you have ever tried to keep a character consistent across separate generations, see our guide to AI video character consistency for why that is usually the hard part — here, the saved persona does that work for you.

How to build a reusable AI spokesperson

1. Define the persona. Set gender, age, ethnicity, style and any additional details. The defaults are gender Young Woman, age 30s and style Professional, so you can generate immediately and adjust from there.
2. Generate the persona image. The face is created with the platform’s image models. Look at it closely before you move on, because this face is the one you will live with for the whole campaign.
3. Save the spokesperson. Saving stores the persona in your browser under a brand-spokesperson key, so the same face returns every time you open the tool.
4. Write the script. Type exactly what the presenter says, in their voice. Keep it short and spoken; short sentences are what a presenter can actually deliver to camera.
5. Set accent and tone. Choose an accent such as American, British or Australian, and a tone or vibe — the default is Confident.
6. Pick your video settings. Choose the video model (the default is ByteDance Seedance 2), the duration, the quality and the aspect ratio (the default is 9:16).
7. Generate. The video renders in roughly 2 to 4 minutes. Watch the first result before you batch anything.

Example scripts for a spokesperson

A spokesperson script is not a blog post read aloud. It is a spoken message with one idea per sentence. These are written to be said, not read.

• Product intro: “Meet the new Trailpack 30. It carries a laptop, a water bottle and a change of clothes, and it still fits under a plane seat. We made it for people who move a lot.”
• Feature highlight: “Setting up takes about thirty seconds. Open the app, hold your phone over the box, and it does the rest. No account, no forms.”
• Trust / service: “Every order ships from our own warehouse. If something arrives damaged, we replace it the same week. That is the whole promise.”
• Onboarding welcome: “Welcome aboard. In the next minute I will show you the three buttons you actually need, and nothing else.”
• Seasonal promo: “For the next seven days, every plan includes the add-on that normally costs extra. Same features, same support, lower price until Friday.”
• Course module intro: “In this lesson we cover the one habit that separates people who finish from people who quit. It takes two minutes to learn and a week to feel.”

Which model to pick

The tool defaults to ByteDance Seedance 2, and for a talking-to-camera spokesperson that default is the sensible starting point. It runs 4 to 15 seconds, supports six aspect ratios including 9:16, and gives you native audio at no extra charge — which matters here, because a spokesperson needs a voice, not a silent clip. Prices are charged per second of video, so length and resolution drive the bill.

Seedance 2 quality	Approximate price
480p	about 7.75 credits per second
720p	about 16.5 credits per second
1080p	about 40.8 credits per second
4K	about 83.2 credits per second

To make that concrete: a 4-second 4K clip is 333 credits. A 4-second 480p clip is a fraction of that. The honest advice is to shoot your first test in a low resolution, confirm the face, accent and tone are right, and only then re-render the keeper at higher quality. The exact price is shown before every generation, and failed steps are refunded. More on how the numbers work is in AI video credits explained, and plan and pack details are on the pricing page. Full settings for the default model are listed on the Seedance 2 model page.

Duration, quality and aspect ratio

These three settings decide what the clip is useful for. Duration is how much script the presenter can deliver — on the default model that is 4 to 15 seconds, so plan two or three short sentences, not a speech. Quality is about where the video will run: 480p and 720p are plenty for a social feed on a phone, while 1080p and 4K are for a landing page, a big screen or footage you intend to crop later. Aspect ratio is about the destination: the default 9:16 suits Reels, Shorts and TikTok, while 16:9 suits a website hero or an embedded ad. Decide the destination first, then set the other two to match, because getting it wrong means paying to render the clip again.

Common mistakes

• Rewriting the persona mid-campaign. Changing age, ethnicity or style breaks the one thing that makes a spokesperson valuable — the same recognisable face.
• Writing a paragraph where a sentence belongs. A 15-second clip holds two or three short sentences, not a full script. Cut until it fits.
• Fighting the accent with the words. A script full of idioms can land oddly in an accent it was not written for. Keep language plain enough to travel.
• Making the persona claim expertise. A synthetic presenter should not announce credentials, qualifications or years of experience it does not have.
• Choosing 4K for a 9:16 social clip. The extra resolution is largely wasted on a phone screen. Match the quality to where the video will actually run.

Disclosure and honesty

This is the part that matters most. A synthetic presenter is a legitimate production tool, but only when you are straight about it.

• Label synthetic presenters where required. Advertising platforms and some jurisdictions expect viewers to know when a person on screen is AI-generated. Where a rule applies, follow it.
• Do not invent credentials. Do not give the persona a medical degree, a law licence, a financial certification or any qualification it does not have. Do not give it a fake real-world identity — a made-up name presented as a real employee or a real expert.
• A saved persona has no lived experience. It has not used your product, visited your store or met your customers, so never have it claim personal experience it cannot have had.
• It cannot appear at live events. A saved persona exists only inside the videos you generate. It cannot take a call, host a webinar or show up anywhere in person, so plan around that.

Used openly, a brand spokesperson is a clean way to keep a consistent on-camera presence. Check your output carries the commercial rights you need before you run it as an ad; our notes on AI video commercial rights cover that. If you also want a narration track without a face on screen, the text to speech tool pairs well with these clips.

Frequently asked questions

Do I have to rebuild the spokesperson each time?

No. Once you save it, the persona is stored in your browser under a brand-spokesperson key, and the same face returns for every future video in the tool.

Which model does the tool use by default?

ByteDance Seedance 2. It runs 4 to 15 seconds, supports six aspect ratios including 9:16, and includes native audio at no extra charge.

How long does generation take?

About 2 to 4 minutes per video.

Can I change the accent and tone?

Yes. You can pick an accent such as American, British or Australian and a tone or vibe; the default tone is Confident.

Should I tell viewers it is AI?

Yes, wherever a platform or jurisdiction requires it. Keep the persona honest too: no invented credentials and no claims of personal experience it cannot have had.

Open the Spokesperson tool to build your first persona: set gender, age, ethnicity and style, generate the face, and save it so it returns every time. New accounts start with 10 free credits and no card required, so you can generate one short 9:16 test and hear how your script sounds in the presenter’s voice before you build a full series.