Does dictation detect the language automatically?

Some dictation tools detect the language automatically, but they usually choose one language for a whole request or recording. They do not switch for each word. Gboard's advanced voice typing is the clearest consumer example, and it works only in a short list of supported languages. The built-in dictation pages from Apple and Microsoft that we read describe choosing the language yourself, not detecting it.

Our guide to multilingual dictation explains the options by platform, and the languages overview covers the wider subject.

Gboard detects the language by default in six languages

Google's Gboard help says automatic language detection is turned on by default for advanced voice typing ¹. The supported list is short: English, French, German (marked "coming soon"), Italian, Japanese and Spanish ².

Google's article ties the feature to Pixel phones, and the device rule differs between versions of the article we found, so check the current version for your phone.

Cloud services pick from a list you supply

Developers who build with a cloud speech service can ask the service to detect the language, but they supply the list of candidates.

Google's automatic language recognition accepts up to three languages ³. Azure's at-start detection accepts up to four candidate languages ⁴. Azure returns one of the languages you supplied, even when none of them is spoken in the audio ⁵. So a list that includes a language nobody speaks can produce a wrong label.

Whisper detects the language from the opening audio only

The open-source Whisper family works at file level. When no language is set, its transcribe function detects the spoken language from up to the first 30 seconds of audio ⁶. That detection is not repeated later in the file, so a recording that changes language after the first 30 seconds can be decoded in the wrong language unless you split it.

Apple and Microsoft describe choosing the language

On a Mac, Apple says you can change the language you dictate in, or set up Dictation for multiple languages and switch between them as needed ⁷. We found no Apple page that describes Dictation detecting the spoken language by itself. On an iPhone, Dictation follows the keyboard you are using.

On Windows 11, Microsoft says that to switch voice typing languages, you change the input language you use ⁸. The article describes manual switching and does not describe automatic detection.

What to do

1. Find out whether your tool detects language. For Gboard, check the advanced voice typing setting on your phone. For Mac, iPhone and Windows, assume you choose the language.
2. Keep any candidate list short. Use only the languages you actually speak.
3. If you run Whisper yourself, set the language rather than letting it detect one, and split recordings that change language.
4. Test a few recordings that match your speech, and count the wrong-language results.

For the cases where detection does not help, see can you dictate in two languages at the same time and how to switch dictation languages quickly. For a wrong language that you did not expect, see why Mac dictation types in the wrong language.