AI Narration: Edward Herrmann vs “Edward”

Curious about AI voices and narration? Here’s one great way to compare human and AI.

Edward Herrmann was a prolific actor and narrator. Even if you don’t recognize his name, you’ll recognize his voice. Now, DeepZen has created an AI voice, “Edward,” using existing recordings of Herrmann from he was alive.

Compare these two short samples:

Listen to “Edward” narrate a sample from the upcoming audiobook “The War With Mexico”

Listen to Edward Herrmann narrate a sample of “Thomas Jefferson: The Art of Power”

What do you think?

To me, “Edward” is a pretty good robot, but a robot nonetheless. Its rhythm, emphasis, intonation, and inflection are algorithmic, not organic. There is no emotional or empathetic reason for its patterns. It has no way to smile or frown as appropriate to match the text. It also has no way to distinguish speakers while performing dialog. I doubt that I could enjoy a whole book read by this pretty good robot.

By comparison, the real Edward Herrmann uses rhythm, emphasis, intonation, and inflection to communicate the meaning, emotion, and subtext of the writing. I hear the person in the recording.

I know we’ll never put the genie of automated voices back in the AI bottle. Nor should we. Quality AI voices like “Edward” do have a place. For instance, they can make written news accessible in real time to people who can’t stop and read it themselves, people with vision or language or time constraints who still need the information.

But let’s keep organic storytelling where it belongs—with the humans. This goes doubly for publishers who can afford the reasonable and declining amounts paid to real narrators, but use robots instead to satisfy their accountants but not their audiences.