Speaking produces more material, faster, than typing. It also produces messier material, which is fine if you structure it afterwards.
Do now: Dictate one task you would normally type, and compare the two.
Say numbers and names slowly and twice. Check the transcript before acting on it.Voice removes the two biggest barriers to these tools: keyboard confidence and a device with a proper keyboard. For many learners, in many languages, and for anyone whose hands are busy or whose sight is limited, voice is not a convenience feature — it is the access route.
Where voice genuinely wins
Volume and speed. Most people speak three to four times faster than they type, and produce more detail because there is less friction between thinking and recording. For capturing what happened, voice is simply better.
Accessibility. For anyone with limited literacy in the written form of their language, limited sight, a hand injury, or no comfortable keyboard, voice is the difference between using these tools and not. This is a substantial share of any real audience.
Situation. Hands full, walking, driving as a passenger, standing in a field, holding a child. A great deal of useful information is generated in exactly these moments and lost because writing it down was impossible.
Voice is not a convenience feature for a large share of learners. It is the way in.
Where it does not
Precision work. Anything with exact figures, references, code, or careful legal or technical wording is better typed and checked visually. Voice is for capture; precision is for the keyboard.
Privacy. Speaking aloud is public. Dictating a member's details in a shared office, a matatu or a clinic waiting room discloses them to everyone present, regardless of how secure the tool is. Think about the room, not only the software.
Accuracy across languages and accents. Recognition remains uneven, and code-switching mid-sentence — which is how a great many people actually speak — is handled poorly. Expect to correct, and check the transcript before acting on anything that matters.
Working well by voice
Speak structure aloud. 'New paragraph.' 'Point one.' 'Question, unresolved.' You are giving the later processing something to organise around, and it costs nothing.
Say numbers, names and references slowly and twice. These are exactly what recognition gets wrong, and repetition gives you a second chance to catch it rather than discovering the error downstream.
Then read the output rather than only listening to it. Spoken responses are easy to accept because listening is passive; errors that would leap off a page slide past the ear. For anything you will act on, look at the text.
Capture by voice, verify by eye. Listening is passive, and errors travel well through it.