Google NotebookLM Expands Multi-Language Capabilities, Including Chinese, and Launches Mobile Apps

Google's AI note-taking tool, NotebookLM, has recently received an update, and its path is becoming clearer and clearer as it moves towards becoming a major personal knowledge management and productivity tool. The latest developments include audio broadcast support for multiple languages, including Chinese, and an upcoming mobile app.

Since its release last year, NotebookLM has gained attention for its ability to transform user-uploaded documents, web links or text notes into structured information. Its feature set includes content summarization, automatic timeline generation and interactive mind maps.

One of the most compelling features of the tool, however, is Audio Overviews. This feature allows users to upload material from a variety of sources (such as text, web pages, PDFs, and even video recordings), which NotebookLM converts into a simulated audio recording of a podcast conversation. The audio typically includes a simulated host asking questions and providing guidance, and "guests" representing different information points of view answering and discussing, creating an interactive and emotional listening experience. This feature is said to be very popular among NotebookLM users.

In previous versions, the Audio Overviews feature was primarily supported in English, which limited its usefulness to the Chinese-speaking user community. Recently, Google announced that Audio Overviews has been expanded to support more than 50 languages, including Chinese (both Simplified and Traditional). To enable the feature, users just need to select Chinese in the "Output Language" option in the Settings menu.

Select the output language in the settings to generate audio in the corresponding language.

In order to check the actual effectiveness of the Chinese audio function, a simple test can be conducted. For example, upload a BBC video recording of Oscar-winning director Zhao Ting in English, and some Chinese interviews with Chinese director Shao Yihui.

After a few minutes of processing, the system was able to generate an 8-minute podcast in Chinese. At the content level, the AI effectively integrates and compares the profiles of the two directors, summarizing Zhao Ting's experience in Hollywood, her creative philosophy, and Shao Yihui's exploration of women's themes and gender relations, as well as his way of working. The voices of the podcast's AI hosts (usually set to be a man and a woman) sound relatively natural and smooth, with natural pauses and colloquial expressions, making the dialogues feel more realistic. The quality of speech synthesis is more vivid than some basic text-to-speech (TTS) services, such as WeChat Book's "Listen to Book" function.

This multi-language capability means that users can handle information sources from different languages. For example, users who do not speak Japanese can quickly grasp the core content by uploading a Japanese news story about Hayao Miyazaki and using NotebookLM to generate a Chinese audio podcast.

Processing information from different language sources and generating audio in the specified language

The core strength of NotebookLM is the way it handles information. The tool is currently based on Google's Gemini 1.5 Flash model, a lightweight but powerful model that supports context windows with up to 200MB of uploaded files and about 500,000 tokens.

Crucially, NotebookLM's design philosophy emphasizes source-grounded answers. This means that its answers and generated content are strictly limited to the information uploaded by the user and can be easily verified by providing clear citations of sources. This feature makes it more useful for users who need precise information sorting and knowledge management, in contrast to many large language model assistants that may "hallucinate" or fabricate information. It is designed to assist the user in understanding and organizing information, not to replace the user's thought process or provide unverified content.

Currently, NotebookLM is primarily accessed through a web browser. While it can be opened in mobile device browsers, the user experience is not ideal. To address this issue, Google has confirmed that it is developing a native mobile app for NotebookLM. The Android version of the app is available for pre-registration in the Google Play Store, while the iOS version is available for pre-registration in the App Store. Both apps are expected to be officially launched at the upcoming Google I/O conference, which is expected to be held around May 20th.

The launch of the mobile app will significantly enhance the accessibility and convenience of NotebookLM, allowing users to organize notes, generate summaries, or listen to audio overviews on the go. The mobile app will reportedly include most of the functionality of the web version and may add new features for conversational interactions with AI presenters.

However, it's worth noting that not all features will be available for free. Premium features like the audio overview may in the future require users to subscribe to the Gemini Advanced service (currently priced at $20 per month) for unlimited access. Free users may be able to get a limited number of trials, but this may limit its popularity as a tool for heavy daily use. The exact functionality and pricing strategy for the mobile app is yet to be clarified by Google when it is officially released.