Google NotebookLM now lets you convert documents into audio discussions

Google NotebookLM is getting a new feature that can turn your documents into engaging podcast-like audio discussions. The Mountain View tech giant announced the new feature, called Audio Overviews, on Wednesday. The feature is available to all users who give permission to try out NotebookLM, which is still available as an experimental tool. The feature isn’t a simple text-to-speech rendering, but uses AI to generate a full-fledged conversation, jokes included. Notably, the tool first rolled out in India and other select regions in June.

Google NotebookLM gets audio summary feature

In a blog afterGoogle provided details about the new feature, noting that Audio Overviews is now rolling out. Gadgets 360 employees spotted the feature, and it should be available to anyone with access to the AI-powered note-taking tool.

The feature is easy to use. The user first needs to add a resource, which can be a PDF, TXT, or markdown file, as well as simple text pasted directly into the notebook platform. Once the resource is added, users need to click on the Notebook Guide icon at the bottom. An interface will appear where the document summary and suggested prompts can be viewed.

On the top right, there’s a new section for Audio Overviews. Tapping Generate will start the process of turning the content into an engaging audio discussion. The format features a male and female AI host discussing the topic.

In our testing, we found the conversation to be very human-like, with the AI ​​hosts highlighting relevant words, modulating voices, and pausing for dramatic effect. Additionally, the AI ​​hosts also interrupt each other to make it seem like a real conversation. The interruptions are done to build on what has been said or to further contextualize the information. The AI ​​hosts also make jokes and banter. Once generated, the audio files are also available for download.

Furthermore, it appears that the feature is using the internet to find more information on the topic, as we sometimes saw the Audio Overview go into more depth than the source material. However, Google states, “It’s important to remember that these generated discussions are not a comprehensive or objective view of a topic, but simply a reflection of the resources you’ve uploaded.”

There are also some limitations to the feature. Google calls Audio Overviews an experimental feature and says that it can take several minutes to generate a single audio discussion. The feature currently only supports the English language, so users cannot generate audio in other languages. Furthermore, the tech giant emphasized that the AI ​​hosts cannot be interrupted and that there may be inaccuracies.