NotebookLM, Google's AI-powered personalized research assistant, first went viral for its Audio Overviews feature. If you aren't familiar with it, Audio Overviews is Google's take on AI-generated podcast-style discussions. With just one click, it creates a conversation between a male and a female host. These discussions have everything you'd expect from a good podcast: witty banter, an engaging flow, and insights that actually stick.
While there are plenty of tools that replicate other NotebookLM features, like its Mind Maps or source-grounded summaries (Gistr is a great example), none really manage to nail the Audio Overviews feature. I tested some of NotebookLM's top competitors hoping to find one, and for the most part, I was disappointed.
The only tool that truly comes close, and actually fixes my biggest gripes with NotebookLM's Audio Overviews, is another Google Labs experiment called Illuminate. But a couple of days ago, Speechify, a tool best known for its text-to-speech capabilities, entered the game with its own AI podcast feature. Of course, I couldn't stop myself from trying it out to see if it could finally beat NotebookLM at its own game.
What does Speechify's AI podcast feature do?
Unfortunately, it's only available on iOS... for now
Speechify launched its AI Podcasts feature on the 12th of August as a part of its back-to-school 2025 release, which it described as one of its "biggest new launches" and it's meant to be a new way for Speechify users to consume content.
Similar to NotebookLM's Audio Overviews feature, with Speechify's AI Podcasts, you can convert any article or document into a "lecture-style" podcast where 2-3 speakers will discuss the key topics. Speechify explains that the podcasts are typically between five and ten minutes long, and users get the option to choose from different style formats.
AI Podcasts are currently available to all Speechify Premium users on iOS. Though I'm not subscribed to the premium plan, I was able to test the feature out by subscribing for a free trial. That said, the fact that it is a paid feature might be a drawback for some, especially considering that NotebookLM's Audio Overviews are completely free to use. Still, I wanted to see if Speechify's paid approach translated to a noticeably better experience.
So, how does the feature actually perform?
Testing it out for myself
Everything I mentioned above is pretty much Speechify's description of the feature, not mine. So, I spent a couple of days playing around with it myself. To create an AI Podcast, all you need to do is head to Speechify, click the plus (+) button at the bottom, and select AI Podcast. If you want to create a podcast from a document, you'll need to upload it first.
Unlike NotebookLM, where you can't really choose a different format beyond the default conversational style (unless you type a custom prompt), Speechify lets you choose from four different podcast types:
* Podcast: The podcast features two speakers and uses a neutral tone.
* Late Night Show: The podcast features three speakers and includes notes of comedy.
* Debate: The podcast features three speakers and is argumentative in tone.
* Lecture: The podcast features two speakers and is delivered in an educational tone.
Since I was essentially testing how the feature stacks up against NotebookLM, I decided to first generate a podcast using the default Podcast format to keep things comparable. I uploaded a PDF of my silly VS Code extensions article I wrote recently and waited for Speechify to generate the podcast.
Waiting for the podcast to generate was oddly fun too. It included playful visuals and captions, with music playing in the background. It actually reminded me a lot of ElevenLabs' podcast feature. After about a minute, the podcast started playing. As it played, a small text indicator above the player read "generating 50%," showing that new portions of the podcast were still being created in real-time. I really liked this, as it made the experience feel more dynamic.
It also cut down on the wait time before I could actually start listening. Instead of staring at a blank loading screen, I could jump right in while the rest processed in the background. With NotebookLM, you often have to wait minutes before hearing anything since it generates the entire Audio Overview upfront.
The podcast Speechify generated was 6 minute and 6 seconds long and hosted by a man and a woman, similar to NotebookLM's default setup. The result was simply incredible. The podcast was concise, naturally incorporated relevant jokes that didn't feel forced, and the hosts used clever analogies. If you've used Speechify before, you probably know its AI voices are among the most natural-sounding in the industry, and that strength really shines here.
Speechify's AI Podcasts also solve some of my biggest frustrations with NotebookLM's Audio Overviews. I'm the type of person who always needs subtitles, and NotebookLM's podcasts don't provide any transcript or on-screen text while playing, which can be frustrating. Speechify, however, includes real-time captions that sync perfectly with the dialogue, making it far easier to follow along or catch up if you miss something.
Another small but great touch: Speechify lets you save your AI Podcasts directly in the app, neatly organized in your library for future listening. It might sound minor, but it's incredibly convenient. I also tried out the other formats Speechify offers, and each brought its own personality to the content.
Where Speechify's AI Podcast feature still falls short
Unfortunately, it's not perfect
As impressive as I found Speechify's AI Podcast feature, I still think it needs some work. Though NotebookLM's Audio Overviews never give you an in-depth overview of whatever source you've uploaded, I primarily rely on it to study. I upload multiple documents and URLs, and the resulting podcast often tends to be quite long.
Speechify, on the other hand, seems to cap podcasts at ten minutes. NotebookLM currently lets you choose whether you'd like to generate a Short, Default, or Long overview, giving you flexibility based on how deep you want to go. Personally, I find five to ten minutes perfect for quick summaries, but having the option to go longer makes a big difference, especially if you're relying on it for self-study or detailed research.
Additionally, Speechify only seems to let you generate a podcast from one document at a time, which limits its usefulness for anyone trying to synthesize information from multiple sources. As I mentioned above, I tend to upload multiple documents simultaneously, and what draws me toward NotebookLM is how well it surfaces connections and highlights relationships between concepts.
I also find it disappointing that the feature is limited to the iOS app only, especially since I primarily generate NotebookLM Audio Overviews on my laptop. That said, I'm sure Speechify will eventually bring it to other platforms.
Finally, I frankly wish this feature wasn't completely locked behind a paywall. I don't mind paying for premium tools (and I would pay for this too), but only if there weren't free alternatives already this generous. For instance, NotebookLM lets you generate three Audio Overviews per day on the free plan. With Speechify's current restrictions, it's hard to justify a subscription when a free competitor already covers so much ground. I just wish they'd offer at least a limited free tier.
Speechify's AI Podcast feature is definitely promising
Trust me when I say this, I've tried to find a tool that does Audio Overviews as well as NotebookLM, and Speechify's AI Podcasts are likely the closest I've come across. I'm genuinely excited to see how the feature improves. Who knows, it might just be the reason I ditch NotebookLM?