Dictation and recorded voice notes look like the same feature and behave differently. Dictation converts speech to text inside a text field and does not keep the source audio with the note; a recording preserves the audio and may also provide a transcript. The useful question is whether you may need to hear what you actually said. Dictation is faster to scan, while a recording gives you evidence when recognition gets a name, number, or tone wrong.
The difference that matters is what gets thrown away
When you dictate, the audio is an input rather than part of the saved note. It becomes characters in a text field, and the note keeps no source recording for you to replay. If the recognizer heard ‘Baxter’ as ‘back sir’, that is now the text unless you catch it on screen.
A voice note inverts this. The recording is the record and the transcript is an interpretation of it, so a wrong name is an annoyance rather than a permanent corruption. The cost is that you have to open something and listen, which nobody does casually.
- Dictate when the content is short, plain, and you will read it back on the same screen.
- Record when it contains a name, a number, a price, or an address you cannot verify later.
- Record when tone carries meaning: a conversation, a reaction, a decision made out loud.
- Dictate when speaking aloud is appropriate and the room is quiet, because that is where recognition is strongest.
How to dictate on an iPhone
The microphone key sits at the bottom right of the standard keyboard, next to the space bar. Tap it inside any text field and speak; tap it again to stop. There is no separate app to install and nothing to enable beyond Settings → General → Keyboard → Enable Dictation.
Two changes explain why current dictation behaves differently from the feature people remember. Apple added on-device processing for many languages in iOS 15 in 2021 on supported iPhones, so those languages can work without a network connection. In iOS 16 in 2022, Apple also kept the keyboard open during dictation, which means correcting a word no longer ends the session. Availability still varies by language and region, and text dictated into a search box may be sent to the search provider.
Say punctuation out loud: ‘period’, ‘comma’, ‘question mark’, ‘new line’, ‘new paragraph’. Without it you get a single unbroken run of words, which is readable today and unreadable in March.
How to dictate on Android
On Gboard, the microphone key appears at the top of the keyboard, and the same rules about spoken punctuation apply. Google began rolling out an 80 MB on-device speech recognizer to Pixel phones in American English in March 2019; support has expanded since, but ordinary Gboard voice typing still varies by device and language.
On Pixel phones, Assistant voice typing (introduced with the Pixel 6 in 2021) adds automatic punctuation and lets you keep the microphone open between sentences. Samsung's own keyboard has a separate implementation with separate language support, so if dictation behaves differently on a Galaxy than it did on your last phone, the keyboard is usually the reason rather than the phone.
Four habits that decide whether dictation is usable
None of these are about speaking more clearly, which is the advice people usually get and rarely the actual problem.
- Name the subject in the first six words. ‘Baxter contract—the renewal date moved’ survives; ‘so they moved it again’ does not.
- Say the punctuation. It costs two syllables and it is the difference between a note and a wall.
- Keep one thought per dictation. Long sessions accumulate errors you will not proofread.
- Read the last line before you leave the screen. Recognition errors are fluent and plausible, so they are invisible unless you look for them deliberately.
Why speaking is worth the trouble at all
The gap is mechanical rather than stylistic. Conversational speech runs at roughly 130 to 150 words a minute; two-thumb typing on a phone runs at a fraction of that for almost everyone. A thought that takes twenty seconds to say takes over a minute to type, and the difference is exactly the window in which you are willing to capture anything at all.
That is also why dictation fails in a specific way: it is fast enough that you will use it in places where recognition is weak. A car with the window down, a station platform, a kitchen with a tap running—the recognizer will still return a confident, fluent sentence, and it will be wrong in ways that read as if you meant them.
What dictation will not do
It will not handle every proper noun reliably. Surnames, product names, street names, and specialist vocabulary are easy to misrecognize, so check them before leaving the screen. Mixed-language support also varies: newer Pixel voice typing can switch automatically among supported languages, while other keyboards still expect you to choose the input language.
It is also not a meeting or call recorder, and nothing here should be read as suggesting otherwise. Dictation captures what you deliberately say into your own phone. Because the source audio is not retained as part of the note, any important figure that you cannot verify on screen is safer in a recording.
Use dictation for short, plain thoughts you will glance at on the same screen, and record when a name, a number, or a tone matters. Say the punctuation, put the subject in the first six words, and read the last line before you walk away—those three habits fix most of what people dislike about talking to their phone.