GPTAvatar Webversion

Starting the app and full screen
Start the app in Chrome/Edge/Safari (please don't use Firefox, see FAQ). Then, click on “Run tool” to start the app and be patient. Due to the file size, downloading the app in the browser often takes 30 seconds to a minute. After starting the app, there's an icon in the bottom right corner to put the app in full-screen mode.

Watch the Youtube-Video:

To gain a better understanding of GPTAvatar’s capabilities, we recommend watching our detailed YouTube video. This video walks you through the key features of the software, starting with an introduction to the user interface and extending to more advanced functionalities. You’ll learn how to create and customize your own avatar configurations, giving you complete control over the appearance and behavior of your digital companion. The video also demonstrates how to collect and analyze user interaction data, making it an invaluable resource for researchers and educators.

Information about the software

• Web and Android Compatibility: The software runs seamlessly in any modern browser and on Android devices.
• Interactive Editing: Create and edit avatar configurations directly in an online editor.
• User Interaction Data: Collect and analyze log data from user interactions.
• Extended Avatars: Includes two additional 3D characters to choose from.
• Customizable Environments: Easily swap the backgrounds of avatars.
• API Key Integration: Set your own OpenAI, Google, or Eleven Labs API keys with ease.

Documentation and collaboration
Documentation of the software is available here.
Feel free to contact me to collaborate on educational and psychological research projects. I am always happy to work on new projects on this topic :) Max Fink (maximilian.fink@unibw.de)

Mini-FAQ
1) Is this software free and can I use it? Yes, it's free and you can use it for research and teaching. If you use it for research, please cite the paper in which we presented the software briefly: Fink, M. C., Robinson, S. A., & Ertl, B. (2024). AI-based avatars are changing the way we learn and teach: benefits and challenges. Frontiers in Education (9), 1416307.
2) How can I edit the config / the content of the avatars in the editor? 
To edit an avatar's configuration, click on "Own config." You can change the voice (including gender and language) by modifying the following tag: "set_friend_google_voice|en-GB-Neural2-C". The voice codes listed after the | symbol can be found here.

You can also change the avatar's name using the "add_friend|Susan" tag. Additionally, the tags "set_friend_base_prompt", "set_friend_direction_prompt", and "set_friend_advice_prompt" allow you to customize the avatar’s behavior. Simply enter the desired text in the language you want the avatar to use, and be sure to conclude each tag with <END_TEXT>.

Once you've finished editing the config, click "Apply." If the config doesn't load correctly due to an error, you can resolve it by reloading the application.
3) Why does the voice of the avatar stutter? Stuttering mainly seems to occur a) when using Firefox as a browser b) with bad internet connection. Please use Chrome, Edge , Safari (not tested) or Chrome o for better audio quality.
4) How do I get the logfiles? Click on "Get logdata" and then copy the content from the editor . Paste it in Word or an editor like Notepad++ and save the content as .csv file. Open it in Excel and click on split text into different columns and select "comma" as separator. The result should be a nicely formatted excel table.

Developers and licenses

This software is a web-based project based on GPTAvatar, created by Seth Robinson (available on GitHub: https://github.com/SethRobinson/GPTAvatar, MIT license). It is being developed by Maximilian C. Fink at Universität der Bundeswehr München, with valuable input from Prof. Bernhard Ertl. Maximilian C. Fink extends his heartfelt gratitude to Seth Robinson for creating and generously sharing this exceptional software as open source, which has enabled further innovation and development. In the spirit of open collaboration, Maximilian will continue to release as many of his extensions and enhancements as open source whenever possible. Some of the latest extensions can be found here: https://github.com/MCF3389.