This podcast was made by humans
Post Reports
The Washington Post
4.4 • 5.1K Ratings
🗓️ 22 October 2024
⏱️ 32 minutes
🧾️ Download transcript
Summary
In recent months, rapid advances in AI audio technology have made it possible to clone any voice, chat with artificial versions of celebrities like Judi Dench, and make a podcast on any topic in minutes.
Those developments have raised ethical questions, opening the door for election disinformation and scams. And they’ve also raised philosophical questions about what complex constellation of qualities allows us to relate to a voice.
Martine Powers speaks with technology columnist Geoffrey Fowler about how Silicon Valley is attempting to re-create the human voice – and where they’re falling short.
Also mentioned in today’s show: How to spot AI-generated “deepfakes” of Kamala Harris and Donald Trump.
Today’s show was produced by Emma Talkoff with help from Elana Gordon. It was edited by Peter Bresnan with help from Monica Campbell and mixed by Sean Carter.
Subscribe to The Washington Post here.
Transcript
Click on a timestamp to play from that location
| 0:00.0 | So Jeff Fowler, my colleague on the tech team at the post, my understanding is that you have something that you want to play for me. |
| 0:10.0 | Yeah, that's right. |
| 0:12.0 | Hi there, it's Martine. |
| 0:14.0 | Ooh. |
| 0:15.0 | I'm here with tech columnist Jeffrey Fowler to talk about advancements in AI audio. |
| 0:20.0 | What? |
| 0:22.0 | In the last few months, AI audio technology has gotten so good that it's a little scary. |
| 0:29.0 | You can use AI to generate a podcast about any topic you want. |
| 0:33.6 | No, oh my God, Jeff, no, that sounds just like me. |
| 0:41.5 | You can use it to make any voice say anything. |
| 0:45.0 | Ah! |
| 0:47.0 | Who, that, that's, obviously I didn't, I didn't say that, right? |
| 0:51.0 | Like that's not from an episode of Post Reports.'s correct so I made that with software from a company called 11 labs and I should say here that the post also uses 11 labs to make AI read-allowed stories on our website |
| 1:02.1 | but I took it a step further I used less than to make AI read aloud stories on our website. |
| 1:02.6 | But I took it a step further. I used less than five minutes of audio of your voice, |
| 1:07.1 | hosting post reports, to make an AI Martin voice that can say literally anything and speak in literally any language. |
| 1:15.0 | I'm Martin Powers. |
| 1:17.0 | I'm Martin Powers. |
| 1:19.0 | What's your Martin Powers? |
| 1:22.0 | How about Japanese? |
| 1:23.0 | She was a team Paul. |
| 1:25.0 | The AI is sending you a message, Martine. |
... |
Please login to see the full transcript.
Disclaimer: The podcast and artwork embedded on this page are from The Washington Post, and are the property of its owner and not affiliated with or endorsed by Tapesearch.
Generated transcripts are the property of The Washington Post and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.
Copyright © Tapesearch 2026.