Description

How much does audio really affect learning? And does it matter whether that voice is human or AI?

In this episode, Matt sits down with Stephanie Warnoff, Senior Market Researcher at TechSmith, to walk through new research that looks at how audio quality and AI narration influence learning outcomes, perceived professionalism, and retention in instructional video.

The study, conducted by TechSmith, was built around a simple experiment. Participants watched the same 45-second how-to video, with the visuals kept the same each time. The only variable was the audio. Viewers heard one of four versions: high-quality human narration, low-quality human narration, high-quality AI narration, and low-quality AI narration. What the team found was...surprising.

Throughout the conversation, Matt and Stephanie explore what the results mean for teams creating training and instructional content, especially as AI tools become easier to access.

This episode is part 1 of a two-part research series. Stay tuned for episode 2, where Matt and Stephanie take a closer look at the second half of the study, focusing on AI avatars, and how different avatar styles compare with human presenters in real learning contexts.

Learning points from the episode include:

1. 00:00 - 01:19 Intro
2. 01:19 - 02:53 Stephanie's background
3. 02:53 - 04:59 TechSmith’s research on AI voices
4. 04:59 - 06:14 The four audio versions used during the study and how the experiment was structured
5. 06:14 - 07:47 What surprised Stephanie about the AI voices
6. 07:47 - 10:58 Is there a need for disclosure with AI?
7. 10:58 - 15:37 Signals for high quality voices
8. 15:37 - 18:04 Why poor audio quality performed better than high audio quality
9. 18:04 - 20:34 Other considerations for effectiveness of learning through AI
10. 20:34 - 23:08 Future research
11. 23:08 - 26:22 Closing thoughts
12. 26:22 - 27:01 Stephanie's final take
13. 27:01 - 28:00 Outro

Important links and mentions:

1. Connect with Stephanie: research@techsmith.com
2. Learn more about the study in this blog post: https://www.techsmith.com/blog/ai-voices-avatars-in-training-videos/