TR
TrainScription
Local AI transcription. No bot. No cloud. Any app.
Awards
Screenshots
Click to view full sizeAbout TrainScription
What is TrainScription?
TrainScription is a local AI transcription tool delivered as a Chrome extension. It captures microphone audio alongside browser tab audio or full desktop system audio, then transcribes speech in real time using Whisper running directly on the user's device through WebAssembly (transformers.js). Audio processing happens entirely in-browser, in volatile RAM-only chunks — nothing is sent to cloud transcription services, and no bot joins your call as a participant. The product emphasizes privacy and offline operation after the one-time model download, with full transparency about its only external connections (model download and license verification — neither touching audio or transcript content). A key differentiator is the Phonetic Brain (TrainScription Lab): a user-trained vocabulary system. Record a quick sample of any name, acronym, product term, or case name, and the system captures the phonetic variants Whisper tends to mishear — then auto-corrects them to the canonical spelling on every future transcript, before you see it. The trained Brain is portable as a .train.json file. Transcripts are speaker-labeled (ME/THEM, customizable), timestamped, stored locally, and exportable as TXT or SRT. Recovery sessions are saved locally in case of crash or accidental close. Full Desktop Mode extends capture beyond the browser tab to any app making sound on the machine — Teams desktop, Zoom desktop, Discord, Slack calls, media players, and more. Not just browser tabs. Free tier: 15-minute sessions, 3 Brain slots, 3 Recovery sessions, per-segment export. Pro: $9.99 one-time, no subscription — unlimited sessions, Brain slots, and Recovery storage, one-click full export, cross-device access, all future updates included. Built for anyone who talks to people with unusual names or specialized jargon — researchers, lawyers, healthcare workers, GMs, journalists, remote teams — who need accurate, private, local-only transcripts without sending a word to the cloud.
Problem this tool solves
Cloud transcription tools add privacy risks, bots, subscriptions, and jargon errors.
How it solves the problem
Runs Whisper locally, captures audio, and learns vocabulary with a reusable Brain.
Target Audience
Professionals, researchers, interviewers, and frequent meeting participants.
Use Cases
- · Transcribe sensitive meetings locally
- · Capture interviews with custom terms
Main Features
Local Whisper transcriptionDual-channel audio capturePhonetic Brain correctionsDesktop and browser modesTXT and SRT export
Categories
Platforms
Browser ExtensionmacOSWindowsLinux
Pricing
Pricing Type: Freemium
Articles about TrainScription
Jul 11, 2026
The False Divide Between Live Conversations And Recorded Ones
Meeting transcription has traditionally focused on live conversations. But some of the most valuable knowledge we encounter isn't live at all. It comes from webinars, conference talks, interviews, lectures, podcasts, and recordings we've collected over time. This article explores why software often treats these as separate categories, why people rarely do, and how TrainScription approaches them as different paths to the same destination: understanding.
TrainScriptionProductivityStartup & Small BusinessAIJul 09, 2026
The Strange Way Digital Information Never Dies
Physical information naturally disappears. Paper gets thrown away. Whiteboards are erased. Notebooks fill up and are replaced. Digital information behaves very differently. It accumulates almost effortlessly, often without intention or purpose. This article explores why "keeping everything forever" isn't always preservation, why forgetting can be valuable, and how TrainScription approaches the lifecycle of conversations differently.
TrainScriptionProductivityAIMarketing & SalesJul 06, 2026
Why We Remember Conversations By Topic Instead Of Time
Most meeting software files conversations by date because calendars are easy to understand. Human memory doesn't work that way. We tend to remember ideas before timestamps, projects before calendar entries, and people before meeting IDs. This article explores why knowledge naturally organizes itself around meaning instead of chronology, and how TrainScription approaches conversation history from that perspective.
TrainScriptionEducation & LearningProductivityAIMakers
HE
@trainscription64 karma
Analytics
Comments
Add a comment...