How to Talk to Continue by Voice on a Mac
Continue is one of the most popular open-source AI coding assistants, and it lives in a chat box inside your editor. The problem is that every prompt starts with typing. On a Mac you can talk to Continue instead, dictating your requests into that same box, and keep your hands on the important keys.
Key takeaways
- Continue is text-first, so voice comes from a system-wide dictation layer, not the extension itself.
- You dictate the natural-language prompt, not raw syntax, which is what Continue wants anyway.
- On-device dictation keeps your voice on your Mac; pair it with a local Continue model for a fully private loop.
- Most people speak around three to four times faster than they type, so prompting by voice is quick.
Why talk to Continue instead of typing?
Continue works by prompting. You describe what you want in plain language, and it edits code, explains a file, or writes a function. That description is the wordy part of the job, and typing it out slows you down. Speaking it is faster and easier on the wrists, which matters if you spend all day at the keyboard. This is the same shift we cover in how to code by voice on a Mac: you are not dictating brackets and semicolons, you are describing intent and letting the AI write the syntax.
Because Continue can run against a local model through Ollama, developers who care about privacy often pick it precisely so their code stays on their machine. It would be a shame to undo that by piping your voice through a cloud transcriber. Pairing Continue with on-device dictation keeps the whole loop, voice and prompt, on your Mac.
What you need first
The setup has two halves. Continue itself is the AI assistant inside VS Code or a JetBrains IDE. The voice half is a Mac dictation app that types system-wide, meaning it inserts text wherever your cursor sits rather than into one dedicated window. That system-wide behavior is the whole trick: the Continue input box is just another text field, so a dictation app that works everywhere works there too.
Set it up in five steps
Install Continue in your editor
Add the Continue extension to VS Code or your JetBrains IDE and connect it to a model. A local model through Ollama keeps your code on your Mac; a hosted model is fine too if you prefer.
Install an on-device dictation app
Download a Mac dictation app that runs speech recognition locally and types system-wide. BlaBlaType does both, with a three-day trial and no card, so you can test the workflow before deciding.
Pick a comfortable shortcut
Set a hold-to-talk or toggle shortcut you can reach without breaking your flow. This is the key you will press right before speaking a prompt into Continue.
Add code terms to a custom dictionary
Teach the app the library names, product names and jargon you say often. Combined with AI cleanup, this keeps prompts readable so Continue understands them on the first try.
Click the Continue box and speak
Put your cursor in the Continue chat or edit input, hold your shortcut, and describe what you want. The text arrives cleaned up, you glance at it, then press send.
Writing prompts that dictate well
Voice prompts to Continue work best when you talk the way you think. Say the goal, then the constraints: "refactor this function to use async await and add error handling for the network call." You do not need to spell out symbols. The natural-language layer is exactly what Continue is built to parse, and on-device AI cleanup strips the filler words and adds punctuation so your run-on sentence becomes a tidy instruction.
If you worry that dictated prompts sound stiff or generic, that concern is more about writing than coding, and we dug into it in can dictation match my writing tone. For prompts, plain and direct is usually best anyway. The workflow is identical to talking to any chat assistant, which we walk through in how to talk to ChatGPT with your voice on a Mac, right down to reviewing the text before you send it.
Talk to your AI tools, hands-free
Dictate prompts into Continue, ChatGPT and any other app on your Mac. Speech recognition runs on-device, so your voice stays local. No card needed for the trial.
Download for macOSKeeping the whole loop private
The appeal of Continue for privacy-minded developers is that the model can stay local. Voice should follow the same rule. With an on-device dictation app, the speech-to-text model runs on your Mac's own hardware using local Whisper and Parakeet models, so your audio and transcripts never leave the machine. If you want the deeper reasoning on why local matters, we broke it down in is Mac dictation private.
This is different from cloud voice features you may have seen elsewhere. Hosted assistants often stream your microphone to a server, as OpenAI describes for its own voice mode. That can be convenient, but for a coding workflow you deliberately kept local, an on-device dictation layer is the consistent choice. You can see the plans on our pricing page once the trial confirms the setup fits your work.
| Approach | Types into Continue | Voice stays on Mac | Setup effort |
|---|---|---|---|
| On-device dictation app | Yes | Yes | Low |
| Cloud voice dictation | Yes | No | Low |
| Apple Dictation | Yes | Mixed | Built in |
| No voice, type only | Manual | Yes | None |
Frequently asked questions
Does Continue have a built-in voice mode?
Continue is a text-first AI coding assistant, so you type prompts into its chat and edit boxes. To talk to it by voice you add a system-wide Mac dictation app that types your words into whatever field has focus, including the Continue input.
Can I dictate prompts into Continue without sending my voice to the cloud?
Yes. If you use an on-device dictation app like BlaBlaType, your speech is transcribed locally on your Mac and never uploaded. Your Continue prompts are handled by whichever model Continue is configured to use, which can be a local model through Ollama.
Will voice dictation handle code terms and library names?
You dictate the natural-language part of the prompt, not raw syntax. A custom dictionary helps with names, frameworks and jargon, and on-device AI cleanup fixes filler and punctuation, so the prompt lands clean in the Continue box.