Speech-to-text on a Mac: four different tools

Last updated 11 October 2026

The four jobs

Most disappointment comes from using a tool built for one of these to do another: trying to dictate email through a transcription app, or expecting a dictation app to open an MP3.

1. Built-in dictation — start here, it is free

1. System Settings → Keyboard → Dictation Switch it on and choose your language.
2. Read the line under the toggle It tells you whether your language is processed on the Mac. This varies by language and region.
3. Set a shortcut you can press without looking The single biggest factor in whether you keep using it.
4. Click into any text field and press it Speak. Press again to stop.

Costs nothing, works in every app, supports a wide range of languages, and on Apple silicon can run on your Mac. What it cannot do is learn your vocabulary, keep a history, or expand spoken triggers into blocks of text.

2. A system-wide dictation app

Same shape — a shortcut, speak, text at the cursor — with the things the built-in one lacks. Vocabulary that learns how your names sound, snippets, a local history with the audio, and control over how the text is tidied.

The shortlist, including apps we do not sell, is at best dictation apps for Mac. If your main complaint about the built-in one is that it keeps mis-spelling the same names, this is the category that fixes it.

3. In-app dictation

Google Docs has voice typing; Word has a Dictate button. Both are good inside their own application and absent outside it. Word's also requires a Microsoft 365 subscription and a connection.

They are worth using if you write almost entirely in one of those two. The moment you switch to email, the habit breaks. See dictating in Google Docs and dictating in Word.

4. Transcribing files

A different job with a different interface: you have a recording and want a document, with speaker labels, a timeline and export formats. Latency does not matter; accuracy and structure do.

Most dictation apps cannot do this. VV's dictation takes live microphone input, but Pro transcribes audio files dropped into a watched folder or imported into Vaitly Meetings (beta), on your computer. For heavy editing of long recordings, a dedicated tool such as MacWhisper has more. See VV vs MacWhisper.

Choosing in one question each

1. Is there an audio file involved? Yes → transcription tool. No → keep going.
2. Are other people speaking? Yes → meeting notetaker. No → keep going.
3. Do you write in more than one app? Yes → system-wide dictation. No → the in-app feature may be enough.
4. Do you correct the same names every day? Yes → a dictation app with custom vocabulary. No → macOS's free one is fine.

What to expect from accuracy

On a decent microphone in a reasonable room, modern speech-to-text on a Mac gets ordinary English right the overwhelming majority of the time. What it gets wrong is predictable: proper nouns it has never seen, homophones where the meaning is ambiguous, and the last three words when you trail off.

All three are manageable — the first with vocabulary, the second and third with a ten-second read-back. See fixing dictation mistakes.