Anker’s new Soundcore Work AI voice recorder is the size of a coin

Anker is the latest company to announce an AI-powered voice recorder designed to transcribe, summarize, and highlight important moments from a conversation or meeting. Its functionality is similar to what’s offered by devices like the Bee or Plaud NotePin, but the Soundcore Work manages to further shrink the hardware down to an even smaller coin-sized wearable.

Launching in September with a wider global rollout expected later this year, the Work measures just 0.91 inches across — and while that’s less than the diameter of a quarter, the recorder is actually a bit thicker than a coin. It’s priced at $99.99, which includes “basic services,” but Anker will also be offering a $15.99 per month subscription that includes additional features not yet revealed.

The Work is designed to be attached to a lanyard or worn on your clothing using a magnetic clip. Weighing just 10 grams, it’s small and lightweight but still manages up to eight hours of battery life, according to Anker. If that’s not enough, the Work’s proprietary charger has a battery of its own that will keep it powered for up to 32 hours.

Instead of capturing every interaction you have all day long, the Work only starts recording audio through its dual mics when tapped, while a double tap lets you mark important parts of a conversation that are highlighted later in transcripts.

The recorder uses GPT-4.1, and eventually GPT-5.0, to transcribe audio recordings in more than 100 languages. Both the transcripts and audio recordings are available in the Soundcore mobile app and can be searched or exported to other formats, but when a user initiates a transcription and a summarization, the encrypted audio files are sent to the AI service provider’s cloud for processing, and deleted from the cloud afterwards. The audio files are also bounced through the cloud when being shared with someone.

Correction, September 4th: Anker originally said the Soundcore Work only sends audio recordings to the cloud when sharing them, but later clarified that they’re also temporarily sent to the cloud when being transcribed.

0 Comments