Follow ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways:
- The NotebookLM app now offers new customization options.
- The platform is positioned as an AI-powered research assistant.
- More updates can be expected soon, according to the platform.
The NotebookLM mobile app just got a little more customizable.
You can now specify more elements within the Google-owned platform’s Audio Overview tool, which debuted in September of last year. Users can adjust the length of generated responses (shorter, default, or longer — currently available only in English), and customize outputs with “guiding prompts,” according to a Thursday X post. Guiding prompts are already available on desktop.
Also: Meta might be secretly scanning your phone’s camera roll – how to check and turn it off
In an accompanying demo video, a user is shown selecting a shorter response, and adds the prompt: “How are Newton’s three laws of motion demonstrated in the launch of a rocket?” After clicking the underlying “Generate” button, the system quickly outputs a brief AI-generated audio clip featuring the platform’s two signature voices, one higher and one lower, interacting in a podcast-style conversation and discussing the physics of rocket flight.
NotebookLM also wrote in its Thursday X post that buffering time on the app has been reduced by 95%. And while the platform has offered multilingual capabilities since April, users can now adjust their preferred language directly within a customization box in the Audio Overview UI; previously, you’d have to adjust the language settings on your device.
More updates can be expected soon, according to the X post.
The rise of AI voices
Google Labs, the tech giant’s experimental AI division, launched NotebookLM in July 2023. It’s been marketed as an AI-powered research assistant that allows users to gain new perspectives on and take a deeper dive into uploaded content, which can vary from PDFs, Google Docs, and images.
Also: Google reveals how much energy a Gemini query uses – in industry first
Google announced last month that NotebookLM’s catalog of notebooks — internal virtual workspaces containing content focused on specific subjects — was expanding to include material from The Economist and The Atlantic, both of which have content licensing agreements with Google. New notebooks were also developed in collaboration with scientists, nonprofits, and authors, including Shakespeare’s work.
NotebookLM’s voice-generating capabilities have almost uncannily humanlike tones, cadences, and quirks. They can do much more than speak in the robotic tenor of a Siri or an Alexa: “This is the magic…uh, well, the physics of liftoff,” one narrator says in the demo video featured in the Thursday X post.
Such lifelike voices have become a major focal point throughout the AI industry as developers race to build more engaging AI tools. Many companies building text-to-speech technology now offer models that can whisper, laugh, realistically imitate the voices of real people, and perform a variety of other impressive audio stunts. These tools could, among many other uses, provide cheaper alternatives to hiring human voice actors for marketing firms or be used to develop more realistic AI companions or therapists, among other use cases.
Also: Google can read your Docs to you now
Meanwhile, NotebookLM and other platforms are hoping to usher in a new era of AI-assisted research and education. Software company Speechify debuted a feature earlier this month that allows Premium subscribers to transform uploaded content into AI-generated “lecture-style” podcasts, or generate a podcast from a text prompt. The tool could be a useful alternative for students who struggle with reading or who just prefer audio-based learning.