Docs
Get started with SleekVoice
Install SleekVoice on macOS, grant Microphone access, use push-to-talk, and fix the problems that stop text from appearing.
Updated
Or install with Homebrew
brew install --cask sleekvoiceOn this page
- How do I install SleekVoice?
- Which macOS permissions does voice typing need?
- How does push-to-talk work?
- What are per-app modes?
- How do I use the personal dictionary?
- What if the install or the shortcut fails?
- What SleekVoice does not do
- Questions
- Does my voice leave my Mac?
- How much does it cost?
- Where is the source?
SleekVoice is a free, open-source voice typing app for macOS. It transcribes speech with Whisper on your Mac, including offline, and types the result into the app that is focused. You hold a push-to-talk shortcut, speak, and let go. A per-app mode and a personal dictionary are part of the published feature set.
This guide is limited to what the project has actually published, plus the macOS permission screens Apple documents for any app that listens and types. SleekVoice has not published a default key name, a settings map, or which Whisper model size it ships. Where those details matter, use what the app shows after you install it.
How do I install SleekVoice?
You need a Mac. SleekVoice does not publish Windows, Linux, iPhone, or Android builds.
The project’s install command is:
brew install --cask sleekvoice
Homebrew (opens in a new tab) has to be installed first. Open Terminal, paste the command, and let Homebrew finish. If Homebrew says it cannot find the cask, the cask is not in the tap your Mac is using. Do not spend the afternoon inventing a tap name. Download a build from GitHub releases (opens in a new tab) instead, open the disk image or archive that release provides, and move SleekVoice to the Applications folder the way you would any other Mac app.
The first launch may ask macOS to confirm an app from the internet. That prompt is macOS Gatekeeper, not a SleekVoice account. There is no SleekVoice login.
Which macOS permissions does voice typing need?
SleekVoice has to hear you, and it has to be allowed to put text into other apps. Those are two different macOS switches.
Microphone access is required for any app that records audio. Apple’s steps are: Apple menu, System Settings, Privacy & Security, Microphone, then turn access on for the app. The support article is Control access to the microphone on Mac (opens in a new tab). If SleekVoice is missing from the list, launch it once and try the shortcut so macOS has a reason to show the prompt.
Typing into other apps is a separate privilege. Apple’s Accessibility pane is where you allow an app to control the Mac. The article is Allow accessibility apps to access your Mac (opens in a new tab). If macOS shows a prompt when SleekVoice tries to insert text, open System Settings from that prompt and turn the app on. If you previously clicked Deny, the same pane is where you reverse it. This guide does not claim a private permission screen inside SleekVoice, because the project has not published one. Use the system prompts and System Settings.
You can revoke either permission later. Voice typing will stop hearing you, or stop inserting text, until you turn it back on. That is the point of the switch.
How does push-to-talk work?
Push-to-talk means the shortcut is held while you speak. Press and hold, talk, release. The transcript is then typed into the focused app. SleekVoice’s own description of the flow is three steps: hold the shortcut, speak while Whisper runs locally, and see the text appear where the cursor already is. There is no separate window you are expected to copy from.
The project has not published the default key combination. Look at the shortcut the app displays, and use that. If another utility already owns those keys, change one of them. macOS lists conflicting shortcuts in System Settings, Keyboard, Keyboard Shortcuts. If SleekVoice offers its own shortcut control, pick a combination that list is not already using. Do not assume a particular modifier just because another dictation app uses it.
Put the cursor in a real text field first. A Finder window, the Desktop, or a button that does not accept typing has nowhere for the words to go. Click into a note, a message, or a document, hold the shortcut, and speak a short sentence you can check.
What are per-app modes?
SleekVoice keeps a mode for each app you dictate into. The project has not published the names of those modes, so this page will not invent “email mode” or “code mode” as if they were labeled that way in the product.
The reason the feature exists is practical. A sentence to a colleague and a line you are inserting into a file are not the same kind of text, and the focused app is the clue. Set the mode in the app for the programs you actually use, then switch apps and dictate again. The mode should follow the focused app rather than forcing you to paste from a side panel.
If a mode seems to stick when you change apps, check that the new app is really focused. A background window can still look selected. Click its text field and try a short phrase.
How do I use the personal dictionary?
The personal dictionary is for the names and words you use that a general speech model is likely to miss: a colleague’s name, a product name, an acronym, a street. Add the spelling you want. The project has not published the button label or whether additions are automatic, so use the dictionary control the app shows.
Say the word again after you add it, in the app where you needed it. If the old spelling is still inserted, check that you saved the entry and that you are dictating into the same Mac user account. The dictionary is part of the local app. It is not a SleekVoice cloud account, because there is not one.
Wispr Flow documents its own dictionary behavior, including an optional auto-add setting, on Wispr’s site. That setting is not a description of SleekVoice.
What if the install or the shortcut fails?
Work down the list. Stop when text appears.
- Confirm the cursor is in a field that accepts typing.
- Confirm Microphone access is on for SleekVoice in System Settings. If the Mac is using an external microphone, select that input in System Settings where the input device is chosen. Apple’s Dictation article describes a microphone-source menu for Apple’s own Dictation; the same idea applies if SleekVoice is listening to the wrong device. Use the input the app or macOS shows, and do not assume the built-in mic.
- If you can see a listening state but nothing is typed, check Accessibility for SleekVoice and turn it on.
- If the shortcut does nothing, another app has those keys, or the shortcut you remember is not the one SleekVoice is using. Read the shortcut in the app.
- If
brew install --cask sleekvoicefails because the cask is missing, use GitHub releases. - If the Mac is offline, local transcription is what the project says still works. SleekVoice does not publish a cloud fallback. If the app has never been opened while online and it asks for a download, finish that once. This page will not invent a model filename.
- Still stuck: email hello@sleekvoice.app or open a GitHub issue (opens in a new tab) with the macOS version, the app you were typing into, and what you expected.
What SleekVoice does not do
It does not publish a speech-to-text HTTP API. It does not publish an MCP server or a plugin. It is not a meeting bot. It is a Mac dictation app. Price is zero. The license is GPL-3.0.
Apple’s built-in Dictation is a different feature, documented in Dictate messages and documents on Mac (opens in a new tab). Apple says you can check Keyboard settings to see whether general text Dictation is processed on device or sent to Siri servers. That note is about Apple Dictation, not about SleekVoice. SleekVoice’s claim is local Whisper on your Mac.
Questions
Does my voice leave my Mac?
The project says no. Whisper runs locally, including offline. The text you accept into another app can still sync if that app syncs.
How much does it cost?
Nothing. Wispr Flow publishes a free plan and paid plans. SleekVoice does not.
Where is the source?
github.com/SuperAgents-Dev/SleekVoice (opens in a new tab), GPL-3.0. Read whatever the repository contains when you start. This guide does not invent a file tree.