Particle.news
Download on the App Store
3 ARTICLES
·
4w ago

TwinMind Raises $5.7M Seed, Launches Ear-3 Speech Model

The ex‑Google X team is pitching a privacy‑first, always‑listening second brain.

TwinMind
TwinMind has announced new funding.
Twinmind app uses Ear-3 AI for ambient speech capture

Overview

  • TwinMind closed a $5.7 million seed round led by Streamlined Ventures with participation from Sequoia Capital and Stephen Wolfram, valuing the company at $60 million post‑money.
  • Its mobile app runs in the background on iPhone and Android, and a Chrome extension adds browser context, with a native Swift service enabling all‑day passive audio capture on iOS.
  • Founders say the app transcribes on‑device offline for 16–17 hours, deletes audio as it is processed, stores only local text, and does not train models on user data.
  • The new Ear‑3 model supports 140+ languages with a 5.26% word error rate and 3.8% speaker diarization error, runs in the cloud with automatic fallback to the offline Ear‑2, and will be offered via API at $0.23 per hour in the coming weeks.
  • TwinMind reports about 30,000 total users and 15,000 monthly active users across the U.S. and several international markets, and it offers a $15 per month Pro tier with a 2‑million‑token context window.