Google NotebookLM Will Now Let You Turn Documents Into Audio Discussions

Google NotebookLM is receiving a brand new characteristic that may flip your paperwork into partaking podcast-like audio discussions. The Mountain View-based tech big introduced the brand new characteristic dubbed Audio Overviews on Wednesday. The characteristic is out there to all customers who consent to check out NotebookLM which continues to be obtainable as an experimental software. The characteristic just isn’t a easy text-to-speech rendering, but it surely makes use of AI to generate a full-fledged dialog, together with jokes. Notably, the software was first introduced in India and different choose areas in June.

Google NotebookLM Will get Audio Overviews Characteristic

In a weblog post, Google detailed the brand new characteristic and highlighted that Audio Overviews has now been rolled out. Devices 360 workers members had been capable of spot the characteristic, and it must be obtainable to anybody with entry to the AI-powered note-taking software.

The characteristic is straightforward to make use of. The person first wants so as to add a supply, which is usually a PDF, TXT, or markdown file in addition to easy textual content pasted straight into the pocket book platform. As soon as the supply has been added, customers might want to click on the Pocket book information icon positioned on the backside. It’ll pop up an interface the place the abstract of the doc in addition to prompt prompts will be seen.

On the high proper, there’s a new part for Audio Overviews. Tapping on Generate will begin the method of turning the content material into a fascinating audio dialogue. The format features a male and a feminine AI host who focus on the subject.

In our testing, we discovered that the dialog could be very human-like, with the AI hosts emphasising related phrases, modulating voices, and taking pauses for a dramatic impact. Moreover, the AI hosts additionally interrupt one another to make it seem like an actual dialog. The interruptions are performed to both construct on high of what was mentioned or to contextualise the data additional. The AI hosts additionally joke and banter. As soon as generated, the audio recordsdata will also be downloaded.

Additional, it seems that the characteristic makes use of the web to seek out extra details about the subject, as we noticed the Audio Overview containing better depth than the supply materials, at occasions. Nevertheless, Google states, “It is essential to do not forget that these generated discussions should not a complete or goal view of a subject, however merely a mirrored image of the sources that you have uploaded.”

There are some limitations to the characteristic as effectively. Calling Audio Overviews an experimental characteristic, Google says that it could take a number of minutes to generate a single audio dialogue. The characteristic at present solely helps the English language, so customers will be unable to generate audio in different languages. Additional, the tech big highlighted that the AI hosts can’t be interrupted, and there is likely to be inaccuracies.