Google’s NotebookLM Now Lets You Customize Its AI Podcasts Leave a comment

Google simply added a brand new customization device for the viral AI podcasts in its NotebookLM software program. I obtained early entry and examined it out utilizing Franz Kafka’s The Metamorphosis because the supply materials, spending just a few hours producing podcasts concerning the seminal novella—a few of them extra unhinged than others.

Launched by Google Labs in 2023 as an experimental, AI-focused writing device, NotebookLM has been having fun with a resurgence in consumer curiosity since early September, when the builders added an choice to generate podcast-like conversations between two AI voices—one male-sounding and one female-sounding—from uploaded paperwork. Whereas these audio “deep dives” can be utilized for learning and productiveness, lots of the viral clips on-line centered on the leisure issue of asking robotic hosts to debate weird or extremely private supply paperwork, like a LinkedIn profile.

Raiza Martin, who leads the NotebookLM crew within Google Labs, is pumped to offer customers extra management over the content material of those artificial podcasts. “It is the primary characteristic we have heard folks request,” she says. “They need to present a bit little bit of suggestions as to what the deep dive focuses on.” In accordance with Martin, that is the primary replace of many coming down the pipeline.

Nearing the one 12 months anniversary of its full launch, NotebookLM can be dropping the “experimental” tag—an indication it’s not headed in the direction of the perpetual Google graveyard of deserted software program, or a minimum of for now. Martin says this label was eliminated because the crew hit inside milestones for general high quality, consumer retention, and interface requirements. She additionally says customers can now anticipate the next degree of stability from the software program.

The best way to Customise the AI Podcasts

To make an AI podcast utilizing NotebookLM, open up the Google Labs web site and begin a New Pocket book. Then, add any supply paperwork you wish to be used for the audio output. These will be something from information in your pc to YouTube hyperlinks.

Subsequent, while you click on on the Pocket book information, you’ll now see the choice to generate a deep dive in addition to the choice to customise it first. Select Customise and add your immediate for a way you’d just like the AI podcast to come back out. The software program suggests that you simply think about what sections of the sources you’d like highlighted, bigger subjects you need additional explored, or totally different meant audiences who you need the message to succeed in.

One tip Martin shares for attempting out the brand new characteristic is to generate the Audio Overview with out modifications, and whereas you’re listening to this primary iteration, write down any burning questions you’ve or subjects you would like it expanded on. Afterwards, use these notes as a launching pad to create your prompts for NotebookLM and regenerate that AI podcast along with your pursuits in thoughts.

My First Impressions

I uploaded an 80-page file of Kafka’s well-known work of existential literature—in it, the principle character wakes up one morning to seek out that he has was a big bug—to see how the customization will work for NotebookLM customers. The primary Audio Overview it generated, sans immediate customization, was a strong, albeit broad overview of what occurs within the novella, in addition to some dialogue of its key themes. Nothing groundbreaking, however respectable.

Pondering like a nerdy faculty English main, which I undoubtedly was, my first immediate adjustment was for the podcast dialogue to focus extra on themes of alienation and overbearing paperwork discovered within the ebook. With the additional nudge, this output from NotebookLM did an admirable job of zeroing in on these motifs and producing a dialogue that sounded just like what I’ve heard earlier than in faculty lecture rooms. It was a bit meandering, however completely listenable.