Write at the pace of your voice.
A shortcut to dictate in the field where you’re writing. Personal vocabulary and reusable snippets help keep your words your own.
Customizable shortcut · personal vocabulary · snippets
Voice. Writing. Memory.
Write by speaking. Turn meetings into a memory you can return to. It all starts with your voice, on your Mac.
Preparing for launch
Let’s review the proposal by Thursday.
I’ll draft the first version and share it with the team.
proposal
Planning the week
00:42 · AnaLet’s review the proposal by Thursday.
Every word finds a place
A shortcut to dictate in the field where you’re writing. Personal vocabulary and reusable snippets help keep your words your own.
Customizable shortcut · personal vocabulary · snippets
Keep audio and transcripts together, review who spoke, mark moments and return to the point where a decision was made.
Editable utterances · markers · audio in context
Search by person, date or word. Organize your library and export timestamped text for the next step in your work.
Search · filters · TXT, Markdown, SRT and JSON
Your Mac. Your words.
Transcription and speaker separation run on your Mac after model preparation. Your recordings are not sent to the developer for this processing.
The initial model download needs internet. If you enable queries, your AI client may send excerpts to its own provider.
Start with care. Stay in your flow.
Choose a model during initial setup. Spokeleaf shows preparation progress and guides you through the permissions it needs.
Use the shortcut, record a meeting or import a file. Vocabulary, snippets and markers stay within reach.
Review the transcript, find an utterance by timestamp and share only what you choose.
Your meetings, with context
Connect a compatible local assistant through MCP and query the meetings you share. An answer can return to the passage, speaker and moment in the conversation.
Reading is off initially, with permission per meeting. Ordinary web AI apps are not a local connection.
Initial preparation downloads the transcription model. Afterward, transcription works locally. An external AI assistant may need internet according to your chosen client and account.
Speaker separation assigns labels to voices. You can review utterances and edit names. Optional enrollment of your own voice requires an explicit action. Similar or overlapping voices may need correction.
Yes. Assistant reading starts off and each meeting starts private. You enable reading globally and share items individually; you can revoke new queries at any time.
We’re preparing the Mac launch. The selected model is a one-time purchase, paid when downloading. Price, final requirements and availability will be announced at launch.
Spokeleaf
Explore Spokeleaf’s flow for dictation, meetings and assistants.
Explore the getting-started guides