There are two kinds of speech-to-text on a Mac: dictation, where you speak and words appear as you type (built into macOS), and transcription, where existing speech — a recording, a meeting, a video, anything playing — becomes a document. macOS handles the first well out of the box. For the second, TranscribeAnything turns any audio source into timestamped text on-device: microphone, system audio, files, and web links, at up to 80× realtime.
| You want to… | Use |
|---|---|
| Type an email or note by voice, live | macOS built-in Dictation (fn fn or the mic key) |
| Turn a recording, voice memo, or file into text | TranscribeAnything — drop the file on the menu bar |
| Get text of a meeting, call, or webinar | TranscribeAnything — System audio capture, live |
| Caption whatever is playing, in real time | TranscribeAnything — Live captions overlay |
| Record a room or yourself and get a transcript | TranscribeAnything — Record mic |
The distinction matters because dictation engines are tuned for one voice speaking deliberately at a keyboard, while transcription handles natural conversation, multiple speakers, playback audio, and long recordings. Use each for what it's built for.
Apple silicon changed the speech-to-text calculus: the Neural Engine runs a full speech model faster than realtime locally, so uploading audio to a cloud service buys you nothing for most jobs — it just adds a meter, a privacy policy, and a network dependency. On-device transcription is faster for files (no upload), works offline, and keeps recordings on your machine. TranscribeAnything runs Apple's on-device speech engine at up to 80× realtime, supports 10 languages, and costs $19.99/year for unlimited transcription — the first 5 transcriptions (up to 30 minutes each) are free.