Voice typing vs dictation is mostly a product-boundary question. A document editor may ship a microphone button that only works in that editor, in that browser, while that tab is focused. An operating system may offer a dictation key that types into any text field the OS considers compatible. A dedicated dictation app aims to own the trigger and the insertion path so the same gesture works in email, chat, and code. Accuracy still depends on the speech model; the typing metaphor just stresses that output is characters in a field.
When people search for voice typing they often want a free, already-installed button. That is a reasonable start for short phrases. Limits show up as soon as you need custom vocabulary, filler-word cleanup, offline use, or a hotkey that follows you between apps. Dictation software exists because those gaps are common. Neither term implies a particular privacy model: voice typing can be on-device or cloud-backed, and so can dictation. Read where the audio is processed, not only what the button is called.