Gemini Notebook Can Record 1 Hour of Live Audio, but the Real Point Isn't the Recording

According to TestingCatalog, Google is gradually rolling out live recording on the Gemini Notebook mobile app. Gemini Notebook is Google's AI note-taking tool. After this update, the mobile app can record up to 1 hour of live audio, and the recording file can be used directly as a source in Notebook, after which users can ask questions and generate summaries based on it.

Recording is only the entry point; the real challenge comes after recording. If an hour of raw audio isn't segmented cleanly and an index can't be built, when you want to look up who said what at minute 47, the model can only guess. Some users pointed out that the core capabilities are chunking and indexing, so users can precisely hit a specific minute and the system doesn't make things up. Others asked: will citations include timestamps?

The 1-hour limit has also sparked different views. Some users think it's just right for recording a lecture, and enough for a stand-up meeting. But people who frequently attend meetings say most meetings run 90 minutes, and the real decisions often happen in the last half hour; with a one-hour limit, you either have to split it into two segments or cut it off.

Some users added two practical habits. Before recording, state clearly: some U.S. states (such as California) require all participants to consent. After the meeting, while you still remember who said what, immediately ask Notebook to extract decisions, owners, and deadlines.

Another view is quite sharp: one hour is enough for most meetings, but the real test is speaker separation and crosstalk. A transcript that can't tell who said what is only half a source.

The significance of mobile recording is that it extends Notebook from processing existing materials to generating materials on the spot. This direction is more interesting than the one-hour limit itself. Whether one hour is enough depends on what you record with it.

Source: TestingCatalog