Skip to main content

How to Get Started with Interactive Avatars

Here you can find some advanced and detailed instructions for creating an optimal Interactive Avatar.

Emily avatar
Written by Emily
Updated this week

Imagine engaging with a lifelike avatar that responds in real-time, bringing a whole new level of interactivity to virtual conversations. Whether for customer support, live presentations, or personalized sales, HeyGen’s Interactive Avatars can step into any role, speaking and interacting instantly. In this article, we'll show you how these avatars can transform the way you connect with your audience.


Where to Find an Interactive Avatar in HeyGen

To get started, navigate to the Labs tab on the bottom left side of the screen, and choose Interactive Avatar, From there, navigate to Avatars.

Once on this screen, you will be able to toggle between 'My Avatars', 'Public Avatars', and 'Examples'. You can either choose one of the provided public avatars and create a Knowledge Base for it, or you can create a custom avatar from scratch!

🛎️ Note: Users with Free plans have access to all PUBLIC Interactive Avatars included in their subscription; CUSTOM Interactive Avatars are available exclusively to paid users. For more details, see the pricing section below.

Free and Creator users can skip down to the section titled 'How to Use an Interactive Avatar' at this time.


How to Create a Custom Interactive Avatar

To start creating your avatar, you'll need to record footage specifically for this tool following the below guidelines.

⚠️ Important Note: Previous footage used for Video Avatars cannot be reused as the guidelines are unique to Custom Interactive Avatars.


Recording Guidelines

Make sure the footage is continuous, with no cuts or edits.

Total Duration: Minimum of 2 minutes, divided into 3 distinct parts.

  1. Listening (15 seconds) ‼️ This portion is SILENT ‼️
    Purpose: To show active engagement.
    Use facial expressions like raising your eyebrows or smiling to demonstrate that you're attentively listening.

  2. Talking (90 seconds)
    Purpose: To convey your message.
    Speak clearly and confidently, ensuring your message is concise and to the point.

  3. Idling (15 seconds) ‼️ This portion is SILENT ‼️
    Purpose: To show an attentive presence.
    Maintain a neutral expression, occasionally nodding without additional facial expressions.

Key Differences:

  • Listening: Engage with facial expressions to show active listening.

  • Idling: Maintain a neutral demeanor and only nod, reflecting presence without extra expressions.

Important Considerations

Background & Environment

  • Choose a clean, static background or a green screen.

  • Avoid moving objects, changing light sources, or reflections in the background.

  • If using greenscreen, keep it off the floor to reduce color spill.

  • Keep the subject at least 5 feet away from the backdrop to avoid shadows.

Camera Setup & Framing

  • Use a 1920x1080 HD camera or better.

  • Frame the subject chest-up; avoid showing hands unless necessary; No full-body recordings (these are not supported).

  • If hands are included (half- or full-body framing), keep them still and in resting position.

🛎️ Note: Interactive avatars currently do not support 4K output due to latency limitations.

Lighting

  • Use soft, even lighting to avoid harsh shadows.

  • Place two lights on either side of the subject with diffusion for a soft look.

  • Optionally, add a backlight above and behind the head for subject-background separation.

  • Warm up the color temperature to maintain natural skin tones.

  • If shadows appear under hands, add a light below the subject.

Performance Tips

  • Maintain steady eye contact with the camera throughout.

  • Avoid loud breathing, swaying, or unnecessary movement during pauses.

  • Keep jewelry, glasses, and shiny clothing to a minimum to ensure clean background keying.

  • Do two takes:

    • One with minimal movement.

    • One with 4–5 slow, subtle gestures (below shoulder height).

Reminder for Talent

  • Begin with an engaged listening pose.

  • Deliver the message with smooth, steady speech.

  • End with a calm idle position—no sudden movements.

For a visual explainer on how to create the best possible interactive avatar, please watch this short video.

Non-human or 3D Custom Interactive Avatars

Have a 3D character, photo avatar, or non-human face you'd like to turn into an Interactive Avatar? Good news—you can use your own custom non-human characters to create an Interactive Avatar in HeyGen!

The process is simple, but depending on the type of character, a quick manual review may be required to ensure everything works smoothly and meets our guidelines.

  1. Create your character

  2. Produce a video in they HeyGen Editing Studio that’s over 2 minutes long and includes the specific Interactive Avatar footage instructions

  3. Submit the video along with a consent video of a real person

  4. The submission will be rejected automatically at first — don’t worry! Just reach out to our team at [email protected] for approval with the rejected Avatar ID [found in the top right corner under the three dot menu ... ] and a screenshot from the platform where the character was created to confirm it's free for use and doesn’t have an identifiable owner or IP restriction

  5. Once approved, you’re all set to enjoy your new Interactive Avatar!

🛎️ Note: Our avatar technology is optimized for characters with distinct human features—like eyes, a nose, and lips. If your non-human character lacks these, the final avatar may not perform well in terms of realism or lip-sync accuracy.

Processing Time for Custom Interactive Avatars

Please note that your Custom Interactive Avatar will take some time to be ready for use due to the tool’s complexity. To ensure the best possible result, follow the guidelines carefully and as accurately as possible.

Here are the estimated processing times based on your subscription type:

Creator User: 3 to 5 days

Team User: 2 to 3 days

Enterprise User: 24 hours


How to Use An Interactive Avatar

First, you'll need to tell your avatar what to say and how to say it! Since your Interactive Avatar is powered by a Large Language Model (LLM), you can guide it using a prompt, which we refer to as a "Knowledge Base" on HeyGen.

To set it up, simply visit the Interactive Avatar page, select a public or custom avatar, and click the Knowledge button. From there, you can edit or create a new Knowledge Base, providing instructions and information for your avatar.

For in-depth guidance on how to fill your Knowledge Base, please visit the Prompting your Interactive Avatar article.

Users with access to Custom Avatars can also customize their Interactive Avatar by changing its voice and/or renaming it.

Once you've decided on the content for your avatar and are happy with the results, you're ready to test it out! To speak with your avatar, press Chat Now.


Integrating Your Interactive Avatar

For general integrations, we offer a starter project on GitHub to help developers get started with Interactive Avatars from the ground up. Additionally, there’s an NPM package available, making it easy to integrate Interactive Avatar functionality into existing web apps. See below articles for full details:


Interactive Avatar Pricing

Creator Plan: Includes 1 custom Interactive Avatar with subscription. More can be purchased as an add on.


Team Plan: Includes 1 custom Interactive Avatar with subscription. More can be purchased as an add on.

Enterprise Plan: Includes 3 custom Interactive Avatars with subscription, with additional avatars available for $500 each per year.

API Pricing: Details on pricing for Interactive Avatars and more are available in the HeyGen API Plans article.

🛎️ Note: For any subscription type, if a user purchased an Interactive Avatar subscription before October 17th, 2024, they will retain that avatar. After that date, users can purchase an Interactive Avatar through the Avatar Add Ons page.


Troubleshooting Interactive Avatar Streaming API

If you're noticing unexpected API credit drainage, it’s often due to open sessions that haven’t been properly closed. Before we assist with recharging your API credits, please check for the following common issues.

🛎️ Note: The below troubleshooting is for users who have purchased a HeyGen API Subscription and are streaming via API.

Open Sessions Consuming Credits

API credits can be drained by sessions that remain in a "connecting," "connected," or "closing" state. To resolve this:

Missing Session Timeout Logic

If you've disabled our default two-minute session timeout (disableIdleTimeout: true), you must implement your own timeout logic. Without it, inactive sessions may stay open indefinitely, consuming credits.

If you want to disable the idle timeout and use custom session management, only frontend logic will be insufficient for this, you could follow a system flow like this:

Basically, you can send the userId and session_id when starting a session and check if the user is still active on the frontend, and implement logic to keep track sessions.

Session Start: Save userId (generated) and HeyGen session_id to your database with a created_at timestamp.


Frontend Monitoring:


Use the USER_STOP event to detect when the user stops interacting. If inactive for like 5 minutes, notify the backend. (or continuously notify like heartbeat approach)


Check visibilitychange to see if the user leaves the tab. After like 5 minutes, trigger a closure.


Send heartbeats every 30 seconds with userId and session_id to mark the session as active.


Backend Handling:

Track last_active (updated by heartbeats) and status ('active' or 'closed') in the database.


Close sessions with POST /streaming.close when:Frontend reports inactivity (from USER_STOP or visibilitychange).A backend task finds sessions with no heartbeat for like 5+ minutes.


Update the session status to 'closed' after closure.


This type of system flow should help you manage sessions.

Recharging Your Streaming API Credits

Once these issues are addressed, we can proceed with recharging your API credits. Please reach out to [email protected] for assistance with this.


Frequently Asked Questions (FAQ)

I want to add this to my site, but I can’t code. What do I do?

The Interactive Avatar API has resources and documentation that will enable most web developers to add the Interactive Avatar functionality to your website or product. Currently, adding Interactive Avatar functionality to your site requires some programming, as there is no easy “copy-pasting” our Interactive Demo to your site with any customization.

Is there a limit on how many connections can interact with the Interactive avatar on my website at the same time?

Yes—by default, the limit is 100 concurrent connections. If you're an Enterprise user and expect more traffic, we can lift that cap based on your usage needs. Simply reach out to your Enterprise Representative to request an increase.

How do you measure streaming credit usage?
When streaming an interactive avatar, 1 API credit equals 5 minutes of streaming. Here's how usage is calculated:

How do you measure streaming credit usage?
When streaming an interactive avatar, 1 API credit equals 5 minutes of streaming. Credits are measured in 1 second increments with a minimum of 30 seconds per session. For example:

  • A session lasting 30 seconds = 0.1 API credit used.

  • A session lasting 1 minute = 0.2 API credit used.

  • A session lasting 1 minute and 43 seconds = 0.346 API credits used.

It's essential to set an appropriate TIMEOUT limit for your users when embedding the avatar into your site. The default TIMEOUT logic is 2 minutes, meaning if a session starts and isn't manually closed, it will default to 2 minutes, using 0.4 API credit.

I already have a custom GPT or LLM. How do I connect it to the Interactive Avatar?

As mentioned before, the Interactive Avatar acts as the 'body,' while the instructions come from either the HeyGen Knowledge Base or the Streaming API. If you want to use an existing prompt, you can copy and paste the text into the HeyGen Knowledge Base (only text is supported, no file uploads). Alternatively, a developer can write code to send text from your GPT, LLM, or STT directly to the HeyGen Streaming API.

Can the Interactive Avatar speak any language?

There are two parts to this answer:

First, users are not directly communicating with the Interactive Avatar—they’re speaking with ChatGPT or the LLM (Large Language Model) you’ve connected. If that LLM supports multiple languages, the Interactive Avatar will be able to speak those languages as well.

Second, the accent of the Interactive Avatar is determined by the Voice ID you assign. HeyGen offers different Voice IDs from various text-to-speech providers, including ElevenLabs, OpenAI, Azure, and Google. The accent and sound of the voices will vary depending on the provider.

I want my users to be able to speak with the Interactive Avatar out loud. Can I do that?

Yes! You can use HeyGen's Streaming SDK in order to add voice chat functionality to your app or website.

My Interactive Avatar is making weird movements. How can I fix this?

If your Interactive Avatar is making strange movements, it’s likely due to the input footage. Since we can’t adjust already rendered footage, you may need to refilm. When recording footage for an Interactive Avatar, it’s important to:

  • Look directly at the camera

  • Limit hand, arm, and head movements to subtle, ‘micro movements’

  • Maintain a pleasant, relaxed expression

By following these guidelines and all outlined above, you can improve the avatar’s accuracy and avoid unnatural motions.

Can I redo my Interactive Avatar?

We offer one free remake of your Interactive Avatar if you’re not satisfied with the final result. This is the only complimentary remake provided. For more information, please contact [email protected]

Which voice can I use with the Streaming API?

You can use different Voice IDs with Interactive Avatars. To explore the available Voice IDs, visit the AI Voices section on the HeyGen homepage.

Developers can choose from HeyGen’s Public Voices, Private Voices created alongside avatars, or voices sourced from third-party platforms like ElevenLabs.

Can I remove the background from my Custom Interactive Avatar?

You can remove the background from your interactive avatar, though this requires more advanced coding skills than most users have. For additional information, please refer to this link; however, note that our regular support center won’t be able to assist with questions on this topic.

How can I remove the watermark from the Interactive Avatar?

The watermark should be removed from your interactive avatar if you are subscribed to our paid API plans and using the streaming credits purchased under that plan.

So, in general:

  • If you’re using the Interactive Avatar via the HeyGen platform (Share button, hosted link, or HTML embed), the watermark cannot be removed—even with a paid plan.

  • However, if you’re using the Interactive Avatar outside of HeyGen (via API token or SDK integration), then any paid API plan (Pro, Scale, or Enterprise) allows you to remove the watermark from your embedded avatar experience.

If you are seeing the watermark on a paid plan, please reach out to [email protected] for assistance.

Where can I find the Interactive Avatar trial token?

We have removed the token / secret mechanism that was formerly used with the interactive avatar streaming process. This is not required anymore to embed Interactive Avatar on your website. For more information on API plans, please see this API article.

Can I use a Photo Avatar as an Interactive Avatar?

You can create an Interactive Avatar from AI-photo if you provide a video (at least 2 minutes long) that follows the specified guidelines above. To do this, you can use HeyGen or another platform to create a Photo Avatar video and use a consent video featuring yourself. Note that this will trigger a moderation flag, and you’ll need to contact [email protected] for approval.

Did this answer your question?