Dictate
Speak into the app you are already using.
Start from a global hotkey, choose live or final-only delivery, and use clipboard fallback when direct insertion is unavailable.
LocalWhisper
Dictate. Transcribe. Keep control.
A native Mac workbench for dictation, recorded audio, searchable transcripts, and optional text actions through speech and language engines you choose.
No card for the trial · One-time purchase · Native Mac app
Where you are typingText at the cursor
Hey Maya, I’m running about ten minutes late, but I’ll send the revised draft before lunch.
Control Option SpaceSpeakDirect insert or clipboard fallback
Press the global hotkey. Speak. Keep working where you already are.
Three connected workflows
From a quick thought to a working transcript.
Transcribe
Turn recordings into editable documents.
Import supported audio, video, and transcript files, then search, edit, replay, rerun, and export the result from one library.
Transform
Run focused actions through an engine you select.
Clean text, summarize, make chapters, extract action items, translate, or ask questions through Apple Intelligence or a configured service.
Transcript workbench
Keep the recording and the text together.
Import supported audio, video, or transcript files. Edit the result as a document, move through it by time, search the library, preserve earlier transcript revisions, and export the version you need.
Meeting capture
Bring the room into the same workbench.
Meeting capture is a separate, user-started workflow for system audio, with an optional microphone mix. Finished recordings open in the same local transcript library, where speaker labels stay manual and editable.
Processing boundary
Choose where the work runs.
LocalWhisper does not provide a hosted inference service. The engine you select determines whether processing uses Mac capabilities, a service on hardware you control, or an optional endpoint you configured.
01
Apple Speech
Use the speech capabilities available on your Mac, including language assets that macOS may need to install.
02
Whisper, built in
Choose a curated WhisperKit model, download it when needed, and run Core ML inference inside LocalWhisper.
03
Your configured services
Connect whisper.cpp, Parakeet Helper, Ollama, LM Studio, or an optional OpenAI-compatible endpoint you control.
System requirements
What your Mac needs.
These requirements cover the base LocalWhisper experience and everyday use alongside the rest of your Mac apps.
- Mac
M1 or newer
Apple silicon only. Intel Macs are not supported.
- Operating system
macOS 26 or later
Earlier macOS releases are not supported.
- Unified memory
8 GB minimum
16 GB recommended for comfortable use alongside your other apps.
- Available storage
2 GB
Enough room for the app and the base experience.
Saved recordings and optional models use additional space. LocalWhisper shows each model's download size before installation.
One-time purchase
$9
Try transcription, dictation, and meeting capture for 14 days before buying. No card required.
One price. Clear terms.
When the trial ends, you can no longer start a new transcription, dictation, or meeting-capture workflow; existing transcripts remain available to view, edit, search, and export.
- Who can use it
- One person, on Macs they personally use.
- Purchase policy
- Updates released during the first 12 months after purchase are included. Covered versions remain usable afterward.
- Refunds
- Request a refund within 14 days of purchase. Statutory rights are not affected.
- Availability
- The trial download and checkout are available above.
Review the LocalWhisper terms, refund policy, and privacy notice.
Before you start
Know the processing boundary.
Where does processing happen?
It depends on the engine you select. Built-in Whisper runs through Core ML on your Mac. Apple Speech can use the modern on-device path for supported locales; legacy recognition follows Apple’s Speech Recognition permission and service behavior. Local services stay on the host you configure. An optional user-configured endpoint can receive the audio, transcript, prompt, or question needed for the action you request. LocalWhisper does not provide a hosted inference service.
When does LocalWhisper use the network?
A model or language asset may need to be downloaded when you choose it. Requests also occur when you use an optional endpoint or network service that you configured, and legacy Apple recognition follows Apple’s service behavior. The selected provider and its location determine where a configured request goes.
What happens after the trial?
When the trial ends, you can no longer start a new transcription, dictation, or meeting-capture workflow; existing transcripts remain available to view, edit, search, and export.
Does LocalWhisper require an account?
The app does not require a SideQuest account for its transcript library. Any account, credential, or access requirement for an optional service belongs to the provider you choose.