Version 4.2 of Recorder is rolling out today via the Play Store with Speaker labels that “automatically detect and label each speaker when recording.” The release is not yet widely available.

Google explains how “Voice models are stored temporarily on-device until speaker labeling of your transcript is completed, and then deleted (usually within a few minutes).

The speaker text labels will be saved in your transcript, and can be updated by you.”.

People are identified as Speaker 1, Speaker 2, etc., and you can “change the name for each speaker in the recording.” They are also identified in the feed with different name colors and Material You shapes.

Notably, Speaker labels will not work if your “Device is too hot.” .

Speakers are identified in real-time, while Recorder automatically starts a new line with each speaker switch.

After pressing “Edit speaker labels” in the toolbar, you can tap the transcript to “Split from here.” Selecting a speaker lets you rename, create a new one, or assign an existing speaker to the passage. 

As of today, speaker labels do not appear in recorder.google.com, which would be great for uniformity in the future