Diksu Download

Meetings to text, with nobody listening in the cloud

Diksu listens to your microphone and your computer's sound at the same time. A Teams, Zoom or Meet call becomes a text file with what you said and what the others said, speaker by speaker.

14 days with every feature, no card needed.

How it works

  • Choose “Transcribe to a text file” in the Diksu menu, with “Microphone + computer audio” as the source.
  • Start the meeting as usual. Diksu writes each sentence to the file with its time while the meeting goes on.
  • If you like, live captions show at the bottom of the screen.
  • When you stop, Diksu tells the speakers apart and opens the finished file in Documents\Diksu.

Why the microphone and the computer's sound at once

Most transcription services hear either only your microphone or only a recording. Diksu listens to both at once, so what the other participants say is written down too.

It works with any program whose sound comes out of your computer. Nobody has to be invited, no add-on joins the call and the others see no bot.

The audio stays on your computer

Speech recognition and telling the speakers apart happen on your computer. Neither the audio nor the text is sent to us or to anyone else, so Diksu also suits meetings about confidential matters.

Tell the other participants before you start that you are writing the meeting down.

What to do with the text

  • Write the minutes from it.
  • Let a slower, more accurate model go over it, one that listens to whole sentences.
  • Paste it into an AI tool and ask for a summary. You decide what is sent where.

Questions

Does Diksu work with Teams, Zoom and Meet?

Yes. Diksu doesn't connect to the meeting program, it listens to the computer's sound, so it works with any program whose sound comes out of your computer.

Can the other participants see that the meeting is being written down?

No, because no bot joins the meeting. So tell them yourself.

How long does telling the speakers apart take?

About five minutes for an hour-long meeting on an ordinary 4-core computer. It runs in the background and the file opens when it's done.

What if the number of speakers comes out wrong?

Pick the right number under “Speakers in last transcript” in the menu and Diksu tells them apart again.