HeyGen Avatar Video Guide: Script to Presenter Video

HeyGen (heygen.com) produces presenter videos from text. You write or paste a script, choose an avatar (one of HeyGen’s stock presenters or a custom avatar made from your own footage), pick a voice, and HeyGen renders a video of the avatar delivering the lines with synchronised lip movement. It is widely used for training, product explainers, internal updates and localised marketing because the same script can be re-rendered with different voices, languages and avatars without a reshoot.

How a HeyGen avatar video is built

A HeyGen video is made of scenes. Each scene has a script segment, an avatar (or none), a background and optional media such as slides, screenshots, text or video clips. The avatar speaks the segment with the voice you selected. Together the scenes form the video, which you export at the resolution your plan allows.

Three choices determine most of the quality. The script determines whether the video sounds human. The voice determines pace and warmth, and the avatar determines whether the presenter looks right for the audience. Visual polish such as backgrounds and lower thirds matters, but it will not rescue a script that reads like a document.

Writing a script that reads well aloud

HeyGen reads exactly what you write, so the script is a prompt for speech. Written English and spoken English differ, and the gap is where avatar videos go wrong. Practical rules:

• Short sentences. Twelve to eighteen words. If you cannot say it in one breath, split it.
• One idea per sentence. Subordinate clauses sound fine on paper and robotic when voiced.
• Write numbers and abbreviations the way they should be said. Twenty twenty-six, not 2026; A P I or the full phrase, not API, unless the voice handles it well in preview.
• Use punctuation for pacing. Full stops create breaths. Commas create short pauses. Ellipses and dashes are unreliable; use a new sentence instead.
• Read it aloud yourself. Anything you stumble on, the voice will too.
• Open with the point. The first sentence should say what the viewer will get; avatar videos lose people fast.

Preview the audio for each scene before rendering the whole video; a preview costs far less than a re-render.

Choosing avatar and voice

Stock avatars are quickest and are fine for internal content, training and most explainers. Pick one whose framing matches your layout (a full-body avatar needs room; a head-and-shoulders avatar sits well next to slides). Custom avatars, made from footage you record following HeyGen’s instructions, are the right choice when the presenter should be a real team member or a consistent brand face. Consent and rights matter here: only create a custom avatar of a person who has agreed in writing, and keep that record with the project.

For voice, audition a few on the actual opening paragraph rather than on sample text. Listen for how each handles your product names and longest sentence. A slightly slower voice usually sounds more credible than a fast one. If you plan to localise later, note that the voice choice may change per language; see HeyGen translation and lip sync.

Step by step

1. Create a new video and choose landscape, portrait or square to match where it will be shown.
2. Paste the script, then split it into scenes at natural topic breaks, roughly one scene per 20-40 seconds of speech.
3. Choose the avatar and voice; preview the first scene’s audio and adjust wording until it sounds natural.
4. Add visuals per scene: a slide, screenshot or short clip that supports the words, plus a text overlay for key terms.
5. Set background and avatar position so the avatar never covers important visuals.
6. Preview the whole video, fix pronunciation, then generate the final render.

Settings at a glance

Rendering uses credits or minutes that vary by plan. Check the platform’s current pricing page.

Organising a series

Avatar video pays off when you produce many videos the same way. Set that up from the first one:

• Use a template. Save your first finished video as a template with avatar, voice, background and text styles fixed. New episodes then only need a script.
• Use the brand kit. HeyGen’s team and brand features let you store logo, colours and fonts so every video matches without manual styling.
• Keep scripts outside the tool too. Store each script as a text file with the video name, avatar, voice and date. When a stakeholder asks to change one sentence six months later, you will have the source.
• Name videos consistently. Series, episode number, language, version: Onboarding-03-EN-v2.

Try it

Scene 1 script: Welcome to the team. In the next three minutes, I will show you how to set up your account, where to find the weekly plan, and who to ask when you get stuck. Let us start with your account.

Scene 2 script: Open the app and sign in with your work email. You will see a short checklist. Work through it in order. It takes about five minutes.

Avatar: stock, head-and-shoulders, right third of frame. Voice: calm mid-range, default speed. Background: plain brand colour. Scene 2 media: screenshot of the checklist, left two thirds.

Swap the voice for one at a slower speed and preview scene 1 again; the pacing change usually matters more than any visual tweak, and once you have the right voice, save the whole set-up as a template.