Speech to Text for Mac: On-Device, No Cloud, Works in Every App
Your Mac already has speech to text built in, and for a lot of people it is enough. Turn on Dictation, press a key, talk, and words appear. If that is what you came for, the short version is in the next section and the full walkthrough is one click away.
This page is about the other case: when the built-in tool runs out of road. You need a transcript of a recorded interview. You need the software to spell your client's name right. You need to know, with certainty, that a recording of a patient never touched someone else's server. That is where VoicePrivate comes in. It runs its recognition engine entirely on your Mac, works with no network connection, and types into any app you already use.
Two Ways to Do Speech to Text on a Mac
There are really only two paths, and picking between them takes about a minute.
Apple's built-in Dictation. Free, already installed, good for short bursts. Fire it off in a text field and it types what you say. It is a dictation feature, not a transcription tool, so it has no concept of an audio file, a speaker, or an export format.
A dedicated speech-to-text app. Adds the things Dictation was never built to do: transcribing recordings you already have, custom vocabulary for names and jargon, speaker labels, exports, and a privacy model you can actually verify. VoicePrivate is this, with the recognition running on your own hardware.
Most people should try the built-in one first. If it holds up for your work, keep it and save your money. The sections below tell you how to get it running, and exactly where it tends to break.
Turning On the Speech to Text Built Into macOS
The setup is short. Per Apple's own documentation:
- Switch it on. Apple menu > System Settings > Keyboard > Dictation, then turn it on.
- Start dictating. Put the cursor in any text field, then press the Microphone key if your keyboard has one, or the Dictation shortcut you set in those same settings.
- Speak your punctuation. Say the name of the mark, like "exclamation mark". Say "new line" or "new paragraph" to break things up. Emoji work too: say "heart emoji".
- Pick the right microphone. In Dictation settings, use the "Microphone source" pop-up menu. Choosing the wrong input here is the single most common reason dictation appears to do nothing.
- Add a second language. Set up Dictation for more than one language and you can switch between them mid-sentence by clicking the language next to the cursor.
Where Built-In Dictation Stops
Four walls, and you tend to hit them in this order.
It only listens live. There is no way to hand it a recording. If you have an hour of interview audio, a lecture, or a client call sitting on disk, Dictation cannot touch it. That alone sends most people looking for something else.
It does not know your vocabulary. Surnames, drug names, case citations, ticker symbols, internal project names. The built-in engine has no dictionary you can teach, so you fix the same misspelling by hand every single time.
Nothing comes out the other end. No speaker labels, no transcript history, no SRT or VTT or Markdown export. Text lands in the field you were typing into and that is the end of it.
The privacy answer is "it depends". This is the one people get wrong. Apple's position is not a flat yes or no: whether your voice input and transcripts are processed on your device rather than sent to Siri servers varies with your Mac and your dictation language, which is why Apple puts a status indicator in Keyboard settings for you to check. That is fine for a grocery list. If you are dictating anything covered by privilege, PHI, or an NDA, "go read the settings panel again after each macOS update" is not a control you can hand to a compliance officer. We break the whole picture down in macOS speech to text: what it does, what it does not.
On-Device Recognition Means There Is No Network Call
VoicePrivate's engine runs locally. On Apple Silicon it uses the Neural Engine on the M-series chip; on Intel Macs it runs the same models on the CPU. Either way the audio is processed inside the machine.
This is an architectural fact rather than a promise about encryption. During transcription there is no API call, no cloud account, and no usage telemetry going back to a server. What that buys you:
- It works on a plane, in a SCIF, in a basement clinic, or anywhere else the Wi-Fi is absent or blocked
- No speech API sitting underneath you that can change its terms, deprecate a model, or raise prices
- Audio that cannot be subpoenaed from a third-party server, because it was never on one
If you want the reasoning in full, see why on-device processing matters and the Mac voice-to-text privacy guide, which walks through what each major app actually does with your voice data.
Dictating in a Language Other Than English
VoicePrivate handles 99 languages, and the language model runs on device like everything else, so switching languages does not quietly switch on a cloud round-trip. Apple's Dictation also supports multiple languages once you add them in settings, with the caveat above: which of those languages get processed locally is not uniform.
Two practical notes if you work across languages. Set the dictation language before you start rather than mid-flow, because a mismatch between the language you selected and the language you are speaking produces confident, fluent nonsense rather than an obvious error. And if you code-switch inside a single sentence, expect to clean up afterwards regardless of which tool you use. No engine on the market handles that cleanly today.
What VoicePrivate Does
- Live dictation into any app. Types at your cursor in Slack, Notion, Word, Mail, VS Code, Pages, or any macOS text field
- File transcription. Drop in an audio or video file, get a transcript back, processed on your Mac
- Fully offline. After the one-time model download on first launch, no connection is needed again
- 99 languages
- Custom dictionary. Teach it names, acronyms, drug names, case citations, ticker symbols, or any term it keeps getting wrong
- Speaker diarization. Labels who said what in a multi-speaker recording (paid plans)
- AI command mode. Reshape a transcript with plain instructions like summarize or pull out the action items, again on device
- Quick Recall, text shortcuts, and per-app modes. Search everything you have dictated, expand snippets, and set different behavior per application
- Five editions: General, Healthcare, Legal, Finance, Insurance, each with domain vocabulary preloaded
- Exports: TXT, JSON, MD, SRT, VTT
The features page has the complete list. For the dictation side specifically, the voice to text for Mac page goes deeper on real-time mode, per-app configuration, and batch processing, and transcription software for Mac covers the file-based workflow.
How the Mac Speech to Text Options Compare
| Feature | VoicePrivate | macOS Dictation | Dragon for Mac | Cloud tools (Otter, Rev) |
|---|---|---|---|---|
| On-device processing | Yes, always | Varies by language and Mac | Core engine runs locally | No |
| Works with no network | Yes | Not guaranteed | Partial | No |
| Account required | None | Apple ID | Nuance account | Yes |
| Live dictation in any app | Yes | Yes | Yes | No (file upload only) |
| Transcribe existing files | Yes | No | Limited | Yes |
| Custom vocabulary | Yes | No | Yes | Varies |
| Speaker labels | Yes (paid) | No | No | Yes |
| Domain editions | 5 (Healthcare, Legal, Finance, Insurance, General) | No | Legal and Medical add-ons | No |
| Price | $84/yr or $9.99/mo, 14 day money back guarantee | Free | Paid, verify current terms with Nuance | Ongoing subscription |
Head to head against a specific tool: VoicePrivate vs. Apple's built-in Dictation and VoicePrivate vs. Dragon for Mac. If you are shopping around more broadly, start with the best dictation software for Mac, or the swap-out guides for Dragon, Descript, and Fireflies.ai.
Who Actually Needs More Than the Built-In Tool
People handling confidential material. Lawyers, clinicians, therapists, accountants. The Legal, Healthcare, and Finance editions ship with domain vocabulary already loaded, so accuracy on medical codes, citation formats, and financial terms is decent from the first session instead of after weeks of correction. If your work is regulated, secure transcription software for Mac covers what to look for.
Switchers coming from Windows. If you ran Dragon or Windows Speech Recognition and just moved to a Mac, the built-in Dictation will feel thin. This fills the same slot: a real desktop speech app with custom vocabulary, file transcription, and cross-app support.
Anyone with a pile of recordings. Interviews, lectures, calls, voice memos. Dictation cannot open a file. This can.
Writers who draft out loud. Talking is faster than typing for most people. On-device is the only architecture where a third party is not holding a recording of your unfinished work.
What You Need to Run It
VoicePrivate needs macOS 13 Ventura or later. Apple Silicon (M1 or later) is required. Long transcription jobs are fast on Apple Silicon, where the Neural Engine does the heavy lifting and your machine stays usable while a file processes.
macOS only. There is no web app and no mobile app, which is the point: a web app would mean uploading your audio somewhere.
Pricing
Plans are $84 per year or $9.99 per month, and both come with a 14-day money-back guarantee. Teams of two to five run $68 per seat per year on floating seats.
There is no free trial. The 14 day money back guarantee is how you test it: buy a plan, check accuracy against your own voice, your own microphone, and the words you actually use, which is the only benchmark that matters, and ask for a full refund within 14 days if it does not hold up. Full details on the pricing page.
Frequently Asked Questions
Does Mac have built-in speech to text?
Yes. macOS includes Dictation, under Apple menu > System Settings > Keyboard > Dictation. It is free and handles short bursts of typing well. It does not transcribe audio files, label speakers, or export transcripts, and whether your audio is processed on device depends on your Mac and your dictation language.
How do I turn on speech to text on my Mac?
Choose Apple menu > System Settings, click Keyboard in the sidebar, go to Dictation, and turn it on. Then put your cursor in any text field and press the Microphone key or the Dictation shortcut you set in those same settings. The full enable and setup guide has the screenshots.
Does Mac speech to text work offline?
It depends which one you mean. With Apple's Dictation, whether your voice input is processed on device rather than sent to Siri servers varies by language and by Mac, and Apple shows that status in Keyboard settings. VoicePrivate is offline by design: after the one-time model download, dictation and file transcription need no network at all. See offline speech to text on Mac.
Why is speech to text not working on my Mac?
Most of the time it is the microphone source, not the software. Check that Dictation is pointed at the input you are actually speaking into, then check microphone permissions, then check that the dictation language matches the language coming out of your mouth. Our 12 fixes for Mac voice to text walks the rest of the list.
Can I dictate in another language on Mac?
Yes. Add more than one language in Apple's Dictation settings and you can switch between them while dictating by clicking the language shown next to the cursor. VoicePrivate covers 99 languages, all processed on device.
How do I add punctuation when dictating on a Mac?
Say it out loud. Speak the name of the mark, like "exclamation mark" or "question mark", and use "new line" or "new paragraph" to break up the text. Emoji follow the same pattern: say "heart emoji".
Is VoicePrivate free?
No. VoicePrivate is purchase-first and there is no free trial. Plans are $84 per year or $9.99 per month, and teams of two to five are $68 per seat per year. Every plan comes with a 14 day money back guarantee, so you can evaluate it risk free after purchase.
Does VoicePrivate work on Apple Silicon?
Yes, on M1 or later, using the Neural Engine. macOS 13 Ventura or later is required.
Can I use VoicePrivate in any app?
Yes. Live dictation types wherever your cursor already is: Docs, Notion, Word, Mail, Slack, VS Code, or any other text field on macOS.
What is the difference between speech to text and voice to text on Mac?
Nothing. Two names, one thing: turning spoken audio into written text. VoicePrivate does both live dictation and file transcription under either label.