We're pleased to welcome Listen Later, an internet service that turns the text of any article on the web into a custom podcast episode, as our newest sponsor. Just email the URL of the article you want to convert to Listen Later, and it will instantly appear in your personal Listen Later feed in your favorite podcast app.
Created by indie developer Yalim Gargher from Istanbul, Listen Later is a modern product that utilizes AI tools to process and narrate the text of articles. Text cleanup is essential, as many articles contain advertisements, image descriptions, disclaimers, and other supplementary material that get in the way of smooth audio. For example, Listen Later is a smart enough feature to avoid the comments at the end of his TidBITS articles. Then comes the narration. We've come a long way since the days of Apple's text-to-speech synthesis of his Fred voice, and the quality of the audio is amazingly good.
Setup is easy. When you create an account, the main Listen Later URL displays your personal podcast URL, a field for the address you want to associate with your feed, the email address to send to, and available credits. All you have to do now is send an email message to the address with one or more URLs specified in the message body (no subject line required).
I first tested Listen Later in the 6,700-word article “When I Met the Pope” by author Patricia Lockwood. In retrospect, this may not have been the best sample of her text, as Lockwood's prose jumps descriptively from branch to branch.
At the reception, if you'll excuse us, we'll be serving wine from Sting's vineyards at the Lapidary Gallery Vin d'Honneur. 'what did you say? 'I yelled at Hope flying on a piece of glass and she pointed at the pamphlet. Sting was staring out from there with a look of intense ferment on his face. Trudy had her arms around his chest from behind, and they were both wearing great health honey, probably fresh from being trampled for seven hours. I think the idea is to meet the Pope and then take a bite of Sting's grapes. I nod at the pamphlet and Hope nods back conspiratorially and slips it into her bag. That's how it is done. we are doing it. Yes, Vermentino is called “Message in a Bottle.'' The same goes for Sangiovese.
I had to go back and reread the original part to realize that Listen Later's AI narration was getting everything right. Or almost right. I noticed several mistakes throughout the 36-minute audio version of the article. Most of them were homographs like “live” with multiple pronunciations, and the word “Sangiovese” was also cut short in the narration, with the final vowel removed. I too thought it meant “Jesus,” but when I looked at the text, I realized that Lockwood intentionally wrote the word “Jesus.” In subsequent tests, Listen Later's phrases continued to impress, but he sometimes stumbled over abbreviations and numbers, especially his prices, such as $2,599. Perhaps it's because of our style that he omits commas for four-digit prices for brevity.
I don't know Patricia Lockwood, but for all I knew, the first voice I heard on “Listen Later” may have been hers. I quickly learned that wasn't the case, but when I asked Gerger about it, he quickly added a setting that lets you choose from six of her high-quality voices. Another early criticism of my audio quality was that the voices were speaking a hair faster than I would have liked. At first, Gerger didn't think anything could be done about it, but a day later he replied that he had found and added the audio speed settings. In the meantime, I found out that the speed issue was my fault. I listen to Marco Arment's podcast on Overcast. This has his Smart Speed option which shortens the silence. Usually works well for people pausing to breathe, but the AI's narration was so smooth that removing the silence made it sound slightly stuffy.
As you can see in the screenshot above, Listen Later also allows you to translate the text into your preferred language, for an additional fee. It's a good sequel to talking about pricing. Listen Later requires you to pay for API calls to process and narrate your text, along with additional translations. These fees are directly related to the amount of texts you ask Listen Later to process. The average article probably costs between $0.30 and $1.50. Listen Later's pricing page provides cost estimates so you can see how much your article will cost.
Listen Later relies on a pay-as-you-go model rather than a subscription. All new accounts receive a free $2 credit to try the service. After that, you can top up your account anytime. Or, if you don't want to think about it, you can set Ask Later to auto-refill by charging your credit card $5 every time your balance drops below $0.10. This model seems fair and transparent, and subscription naysayers should welcome it.
One thing to remember is that Listen Later, true to its name, is not instantaneous. The length of the article affects the time it takes to convert, as does the load on the OpenAI server. The load is high during US business hours. In the case of Patricia Lockwood's long article, Gerger said the conversion took about 20 minutes to complete, and in some instances the application waited up to 5 minutes for a response from her OpenAI server. However, this normally should not be a problem. The whole point of “listen later” is to queue up articles to listen to, for example, in the car on the way home or during your next workout. Listen Later includes a URL to the original article in the podcast episode description for easy reference.
I discovered that you can also convert PDFs to audio. Andy Weir put it to the test in a recent bonus chapter he wrote. Martian. Additionally, Gerger said Listen Later has added the ability to convert text in images in JPEG, PNG, and WebP formats to audio. I was also skeptical at first. What use cases could there be? It turns out it's common for certain communities to post screenshots of their PDFs on their X/Twitter. So if you're interested, just send the URL to the image to Listen Later for conversion, just like you would for a standard article.
It's funny that I'm writing about Listen Later, which converts text to podcasts, in the same week that Apple introduced the ability to convert podcasts to text. Generative AI has many problems, but some of the features it enables are magic.
If you love listening to podcasts but don't have time to read everything you want to read, try Listen Later with the $2 free credit you get with every new account.
