Flexible voice capture
Hold a shortcut, double-tap for hands-free capture, use a mouse button, or say a configurable wake phrase. Speech recognition runs on the Mac, and recording stops when speaking stops.
Speek
Speek is a macOS voice assistant in the Mac notch for dictation, text editing, answers, and hands-free actions across supported apps and tools.
Voice control and dictation for Mac
Speek is a voice-to-action assistant for macOS that is designed to live in the Mac’s notch. It combines dictation, conversational questions, and requests to operate connected apps or the visible computer interface, so users can speak without switching away from the active application.
A shortcut, hands-free gesture, or wake phrase starts a request. Speek can transcribe into the current text field, rewrite selected text, answer with speech and cards, or carry out actions such as drafting messages, managing calendar items, searching connected services, and controlling media. The project is cloud-connected, supports selectable model providers, and exposes settings for permissions, integrations, voice, dictation, privacy, and memory.
Features
Dictation and voice editing
Insert transcription into the current text field and choose raw, lightly cleaned, or polished writing. Speek can adapt output to the surrounding text and edit selected text by voice.
Conversational assistant
Ask questions, receive spoken replies, continue a conversation, and submit requests for actions. Short requests appear in the notch while longer computer tasks can run in the background and report back.
Context-aware requests
Use screen context, a pointer-circle gesture, pasted or dropped images and files, and selected text to give requests more context. Screen sharing for requests can be turned off.
App and tool integrations
Connect supported Mac apps, MCP servers, plugins, local command-line tools, and skills. Listed connections include Mail, Messages, Calendar, Reminders, Notes, music, Files, Gmail, Google Calendar, GitHub, Notion, and Linear, with additional services available through Composio.
Memory and repeatable workflows
Store facts, preferences, procedures, vocabulary, corrections, and spoken shortcuts. Dictation history, reusable prompts, schedules, and optional recovery of failed recordings are also listed.
Use cases
Hands-free writing
Speak into Mail, Slack, Notes, a browser, or a terminal and insert the resulting text into the active field. Choose a writing mode when you want a raw transcript, cleanup, or a more polished draft.
Editing text in place
Select existing text and ask Speek to shorten, translate, or otherwise revise it, while retaining the captured selection checks described in the project requirements.
Managing everyday Mac tasks
Ask for a calendar lookup, create or change reminders, search notes, draft a message, or control music without manually navigating each app. Consequential actions can appear for review before execution.
Working from visual or file context
Provide screen, pointer, file, image, or PDF context when a question depends on what is visible or attached. Speek can use computer interaction when a connected tool is not available.
Extending workflows for tools and development
Use MCP plugins, local CLI tools, skills, and coding-assistant hooks to route tool-related requests or respond to an existing Claude Code or Codex session by voice. The source says Speek does not start coding tasks itself.
Pros and Cons
Pros
- Combines dictation, text editing, questions, and computer actions in one Mac-focused interface.
- Supports both direct text insertion and richer outputs such as editable email, forecast, and map cards.
- Offers multiple input methods, configurable writing styles, memory, and follow-up conversations.
- Can extend beyond native Mac apps through MCP servers, plugins, skills, and trusted command-line tools.
Cons
- The supplied project materials do not establish packaged distribution details or product-specific pricing.
- Some workflows depend on configured permissions, connected services, selected model providers, or local tools.
- The project documentation distinguishes implemented source paths from live-tested capabilities, so not every listed capability should be treated as production-ready without verification.
FAQ
How do you use Speek?
Speek is a macOS application that runs from the Mac notch. Users can hold a shortcut to speak, enable double-tap hands-free capture, or use voice activation with a chosen wake name. The source also lists permissions for microphone, accessibility, and screen recording where applicable.
Where can Speek put or edit text?
Dictated text can be inserted into the active text field, including in apps such as Mail, Slack, Notes, browsers, and terminals. Speek can also rewrite selected text, with modes ranging from raw transcription to polished writing and configurable style.
What kind of output does Speek provide?
Speek can answer questions aloud, continue a conversation, and return copyable or insertable answers. It can also present cards for workflows such as editable email drafts, weather, and map results.
Does Speek use cloud services?
The repository describes Speek as cloud-connected and lets users choose providers including OpenRouter, OpenAI, or a local Codex login. It also supports local speech recognition and lists optional screen, file, and app context, so the exact data flow depends on the selected model, integrations, and enabled settings.
How much does Speek cost?
The project is a public GitHub repository, but the supplied sources do not establish product-specific pricing, a packaged distribution model, or whether any connected model providers require separate paid accounts.
Quick Facts
- Category
- AI assistant and productivity tool
- Platform
- macOS
- Interface
- Mac notch panel plus main conversation window
- Primary workflows
- Dictation, voice editing, questions, and app or computer actions
- Repository
- github.com/aveekpatra/speek
- Model options
- OpenRouter, OpenAI, or a local Codex login are listed in the repository
Speek Alternatives
SpeakoFlow
SpeakoFlow is a free, open-source desktop voice-to-text app for Windows, macOS, and Linux. Dictate into any app, use Flow, cleanup, translation, and a screen-aware assistant.
Flunkey
Flunkey is a voice-first productivity layer for Windows that turns speech into text, AI outputs, and remembered context without leaving your app.
Sanota
Sanota turns spoken memories and interviews into clear written stories for personal storytelling, family history and shared memories, with guided prompts and subscriptions.
Carbon Voice
Carbon Voice is an async voice messaging app for teams and individuals, with transcripts, AI catch-up, and cross-device access.
Lemon
Lemon is a Mac voice assistant that turns spoken instructions into finished writing tasks and actions. Free Basic and paid Pro plans.
QuickQuill
QuickQuill is a local-first macOS dictation and transcription app to record meetings, summarize audio, and export notes without the cloud.