Voice typing for Slack on Mac.

Voice clips are not the same thing

Slack has voice clips, and they solve a different problem. A clip is audio: the person on the other end has to listen to it, at your pace, in order, and they cannot skim it, search it later or quote one line of it back to you. Sending one is faster for you and slower for everyone else, which is why they are used sparingly by people who are thoughtful about it.

Dictation produces text. It is fast for you and unchanged for everybody else, which is the only version of this trade worth making in a shared channel.

How it works in Slack

your shortcut

Listening… 0:04

Hold it to talk, tap it to toggle, Escape cancels. Pick the key in Settings.

There is nothing to install on the Slack side. Zumbo types into whatever field has the cursor, so every part of Slack is covered without it knowing:

• The message box in a channel or a direct message.
• A thread reply, where the long explanations usually belong.
• The search field, when you are looking for a phrase you half remember.
• Your status, a canvas, a channel description, a reminder you are setting.

1. Put the cursor where the message goes.
2. Hold your shortcut, which you chose in Settings during setup.
3. Talk for as long as you need. There is no session limit.
4. Let go. The text lands in about a fifth of a second, punctuated.
5. Read it, then press return.

The messages this changes

Not the ones that are already short. Nobody needs help typing on my way. What changes is the message you are currently not sending: the context for the person picking up your work tomorrow, the reason the obvious fix will not work, the summary of what happened in the meeting they missed, the answer that deserves three paragraphs and gets one line because you are tired.

Those are the messages where a team either has the information or does not, and they get skipped for a reason that has nothing to do with how important they are.

Getting names right

Slack is full of proper nouns: colleagues, channels, services, customers, the internal tool nobody outside the company has heard of. A general speech model turns all of them into ordinary words. Teach fixes each one permanently in about three seconds, and if you work in software the developer tools vocabulary pack covers most of the rest before you start.

On device, which matters in a work chat

A Slack message can contain a customer name, an incident detail or something about a colleague. Zumbo has no server for your voice: no account, no API key, no network request needed to transcribe, and the speech engine ships inside the app. The finished text is typed into the box and nothing about the audio went anywhere.

Related: dictation for developers, dictating into Notion, and how Zumbo keeps your voice on your Mac.