AYARUXLast updated:
Dictation and translation are two steps
First the phone turns your voice into text; that is speech recognition. Then the text is translated into another language. A word misheard in the first step is mistranslated in the second.
So glance at the text after you speak. Fixing a sentence in your own language before you tap translate is far easier than fixing it in a language you do not know.
Which language should you speak?
Speak your own language. Speech recognition assumes you are speaking the language it is set to; say an English sentence while it is set to Turkish and the result will be odd.
The translation writes the foreign-language text for you. Speaking naturally in your own language improves both the recognition and the translation.
Getting recognised better
Speech recognition works best on short, clear sentences. A few habits that help:
- Hold the phone close to your mouth, and shield the microphone with your hand in wind or crowds.
- Speak sentence by sentence and pause briefly at the end of each.
- Fix names, product codes and addresses by hand after speaking.
- Say numbers in full: “three thirty in the afternoon” rather than “half three”.
Punctuation and capitals
Dictation often leaves punctuation out or adds only some of it. A question mark at the end can change the meaning after translation: “You are coming tomorrow.” and “Are you coming tomorrow?” become different sentences.
On short messages, adding punctuation by hand takes seconds and noticeably improves the translation.
Where does your voice go?
Turning voice into text is the job of your phone's speech recognition service; on some devices it runs on the phone, on others on the provider's servers. The phone's speech recognition settings show which one is used.
Translation is a separate step. If the app you use translates on the phone, the resulting text is not sent anywhere else to be translated.