Press kit — June 2026
Local Whisper transcription that learns your language.
TrainScription is a Chrome extension that transcribes dual-channel audio — your microphone and your browser tab — entirely on-device. No bot joins the call. No audio ever leaves the machine. Whisper AI runs locally via WebAssembly. It works fully offline after a one-time model download.
The differentiator is the Phonetic Brain: a personal vocabulary trainer that captures the phonetic variants Whisper hallucinates for unusual words — names, acronyms, product terms — and auto-corrects every future transcript before the user sees it. Competitors give you Whisper. TrainScription gives you Whisper that already knows your jargon.
In action
Live Studio — real-time dual-channel transcript with Phonetic Brain correction active
Story angles
Privacy / local-first
Zero cloud. No audio transmitted. Works offline after first run. No analytics, no tracking, no server infrastructure.
No meeting bot
Captures tab audio directly — no participant joins the call. No IT flags. No host notification. Invisible to other attendees.
Whisper in the browser
On-device WASM Whisper inference running as a Chrome extension is technically novel. Most Whisper implementations are server-side.
Subscription fatigue
$4.99 once, no subscription, no credit card for the free tier. Otter.ai and Fireflies both charge monthly.
The Phonetic Brain — the angle nobody else has
Every professional has words Whisper gets wrong — a client name, an internal acronym, a product term. The Phonetic Brain captures those mishearings as variants, then silently corrects them on every future session. The trained vocabulary is portable as a .train.json file. It’s personal, it’s calibrated, and it’s the feature competitors can’t easily copy — because it’s yours.
How it compares
Pricing
Free
$0 no credit card, no account
Pro — founding price
$4.99 once, no subscription
Technical
Manifest V3
Chrome extension
WASM Whisper
via transformers.js
Dual-channel
mic + tab audio
3 model tiers
tiny / base / small
Offline-capable
after first download
No server
zero cloud infra
From the developer
“I built this for my own tabletop RPG sessions — I needed something that could hear character names Whisper had never encountered. The Phonetic Brain started as a personal fix. It turned out to be the whole product.”
— Terrance Henry, developer
Contact