Skip to content

What's new

Shipped

Paid passes on the Russian site

Three passes and a five-hour top-up are on sale on the Russian site. Each is paid once, with no auto-renewal.

Shipped

Prompts during negotiations

Select a passage of the transcript in a Work and meetings project, and Wisot suggests what to say now and how to move the deal forward.

Decision

Russian product names

The Russian site now gives the study product and the meetings product Russian names. The English site keeps the name Wisot.

Shipped

Projects for work and meetings

Project settings offer two kinds: Study and Work and meetings. In a meeting, Get a brief answers from the project documents about the last few lines.

Shipped

Accurate recognition

A second recognition pass: the GigaAM v3 model refines each draft phrase a few seconds later. It runs on your computer, and the audio never leaves it. Russian speech only; it needs Chrome, Edge or Yandex Browser, 8 threads and 8 GB of memory.

Shipped

Term dictionary

A list of course terms tells recognition how they are spelled. The dictionary stays in your browser and is never sent to the server.

Shipped

Microphone-only mode

For in-person meetings: Wisot listens to the computer's microphone instead of a tab.

Shipped

Who is speaking

Wisot tells voices apart in real time and labels the lines voice 1, voice 2. Double-click a voice to name it; giving two voices one name merges them. Needs a processor with 4 or more cores.

Shipped

Microphone and Snapshot

The microphone became an optional second source: your lines are labelled you, the tab's audio lecturer. The Snapshot button or Alt+S saves a frame with the slide, and Wisot turns it into text, formulas included.

Shipped

Android: speech recognition on the phone

The Android app recognises speech with the GigaAM v3 model right on the phone.

Decision

Answer by voice or by text

We added text input: it helps in noisy places and suits people who would rather type than speak.

Shipped

Audio question generation prototype

First working version: questions generated from an audio transcript.