At its Made by Google event earlier this year, Google announced that the Recorder app on Pixel devices would soon be able to differentiate different speakers thanks to a new feature. The company demonstrated the feature during the event, showcasing how it used machine learning to automatically identify different speakers and transcribe each speaker's words separately. Google said this new functionality would roll out to the Pixel Recorder app later this year, and it seems like that time is finally here.

9to5Google reports that the new Speaker labels feature will reach users with the latest Pixel Recorder update (version 4.2). The update has already started rolling out via the Play Store, but it's not available widely at the moment. Nonetheless, you should get it on your Pixel in the coming days, after which you'll see a new splash screen in the app explaining the feature.

Screenshots via 9to5Google

Once enabled, the Recorder app will automatically identify and label multiple speakers. It will give each speaker a generic name, but users will have the option to change their names in the recording. Google explains that the app will temporarily store voice models on the device to label speakers. Once it completes labeling speakers in a transcript, it will delete the voice models in a few minutes.

To label different speakers in a recording, the Recorder app:

  • Computes models of the voices detected in the recording,
  • Gives each voice detected a generic text label (such as Speaker 1, Speaker, 2), and
  • Compares the voices throughout the recording so the transcript can reflect what each voice said and when.

As mentioned earlier, the Speaker labels feature is rolling out to users with Pixel Recorder v4.2 The update is not available widely, but you can download the update from APKMirror and sideload it manually. The feature reportedly works as intended when sideloading the update on the Pixel 6 Pro and Pixel 7 Pro running Android 13 QPR1 Beta 3.1.


Via: 9to5Google