Do podcast listeners actually prefer synthetic voices to human ones?

Mindful Notions by Sam H Li

0:00

-9:44

Do podcast listeners actually prefer synthetic voices to human ones?

Who (or what) does it better?

Sam H Li

Dec 29, 2024

Transcript

macro photography of silver and black studio microphone condenser — Cover photo by Jonathan Velasquez on Unsplash

Introduction

Like many people, I witnessed the extraordinary capabilities of Google's NotebookLM. It felt like another fascinating step forward in AI's accessibility to non-engineers. While we've had two years of mainstream ChatGPT, until now AI tools have primarily been text-based for most everyday users.

When I first heard NotebookLM's voices, I wasn't threatened—more contemplative. It's remarkable that such technology exists.

But this raises an important question: Do listeners actually prefer synthetic voices to human ones?

I'm curious where I could find reliable research and credible sources on this topic.

Here is my (quite simple) thought process

Key question: Do people prefer human-voiced podcasts over synthetic-voiced ones?
1. Hypothesis: The preference likely depends on the listener's purpose. NotebookLM's current capabilities are limited to converting existing information into audio format without adding original insights. While this might be useful for personal research consumption, it's not ideal for wider distribution. In contrast, human-made podcasts (even those aided by AI) are typically crafted for specific audiences and offer new perspectives.
Research: Should I use Google or ChatGPT?
1. Decision: I chose Google!…And then ChatGPT
Look into NotebookLM's specific capabilities
Form a conclusion

Googling

The first article that popped up when I Googled was: " Do people prefer to listen to human-voiced podcasts vs. synthetic-voiced podcasts?”

The third search result: Reddit! and it was specifically about NotebookLM

The next step: consulting ChatGPT

“Studies and surveys consistently show that most podcast listeners prefer human-voiced podcasts over synthetic-voiced ones. This preference is driven by several factors:

1. Emotional Connection and Authenticity

Human voices convey emotion, tone, and nuances, making it easier for listeners to connect with the speaker.
Synthetic voices, while improving, often lack the natural rhythm and imperfections that make speech feel genuine.

2. Trust and Credibility

Listeners tend to trust human speakers more, as they feel there's a real person behind the words.
Synthetic voices can sometimes feel impersonal or robotic, potentially reducing the perceived credibility of the content.

3. Listening Experience

Human voices often provide a more engaging and enjoyable listening experience.
While synthetic voices have made great strides in natural-sounding speech, prolonged listening can still feel monotonous or fatiguing to some audiences.

4. Podcast Genre Considerations

For storytelling, interviews, or conversational podcasts, human voices are almost universally preferred.
Synthetic voices may work better in certain niches, such as highly technical content, where personalization isn't critical.

5. Cultural and Accessibility Factors

While many people prefer human voices, synthetic voices play an essential role in accessibility, especially for creating multilingual or text-to-speech content for the visually impaired.”

NotebookLM’s specific capabilities

NotebookLM vs. NotebookLM Plus features

NotebookLM is “an AI-powered research assistant that can help you get to critical insights faster and easier, all grounded in the information you provide.”