A Pendant That Tries to Listen Only to You

Nirva has raised $25 million from investors. That brings its total funding to $33 million, and it raised $8 million earlier this year, according to TechCrunch.
The company was started in 2025 by Wei Lyu, Jason Chen and Hanying Hu. Lyu and Chen previously worked at Meta, in the group that builds virtual and augmented reality products. Earlier backers include Lightspeed and South Park Commons.
The first product is a pendant that can also be worn as a bracelet. It weighs about eight grams, and its battery lasts up to three days on one charge. It sits close to the wearer's mouth, like other small devices worn on the body.
Nirva says the pendant is built to hear only the person wearing it and to tune out the voices of people nearby. To make this work, users read short snippets of their own voice during setup. The device learns that voice from the recordings. Setup cannot be skipped.
By default, Nirva does not keep audio recordings or written transcripts after it processes a conversation.
In my view, there are two separate privacy questions, and the announcement answers only one. Deleting recordings and transcripts limits what is stored after a conversation ends. The voice samples collected during setup are different. They stay in the system as a lasting part of how the device works. How long they are kept, where they are processed and who can reach them are questions that businesses and privacy-minded buyers will want answered before they trust the device.
The broader context here is that devices that record constantly create a problem for everyone nearby, not only the person wearing them. A pendant that tries to block other voices by design changes that picture, at least on paper. The real test is whether it works in a noisy restaurant, on a crowded train or when several people talk at once. The announcement does not yet show that.
If the filtering works as described, the benefit is modest but real. The device could help with taking notes, remembering details or recalling meetings, without capturing people who never agreed to be recorded. That kind of restraint is uncommon in this category, and it is the part worth watching as the product moves from funding news into people's hands.
The $25 million is the news. What the announcement actually sets out is three promises: the device listens only to its wearer, it learns that wearer's voice during setup, and it keeps no audio by default. The company will need to show those promises hold up in everyday use.


