Fact · Languages and accents
94languages
In the languages supported by speech recognition systems dataset, the 10 published figures for “Cloud speech API” have a median of 94 languages, ranging from 13 to 125.
Calculated from 10 data points on October 19, 2026
- What it measures
- Each figure is one row of the dataset: a published measurement for a stated group, traced to its source. The median is the middle value of the 10 figures for this group.
- How it is calculated
- Median of 10 values (13, 18, 55, 60, 90, 98, 99, 100, 100, 125) = 94.
- What it does not show
- Makers count differently: some list languages, others locales. Support also varies by device and feature, so a language on the list may not work offline or everywhere.
A statistic of the dataset How Many Languages Speech Recognition Supports, calculated by code from its published rows.
Calculated from
- AssemblyAI's documentation says its Universal-2 speech-to-text model supports 99 languages.
- AssemblyAI's documentation says its Universal-3.5 Pro speech-to-text model supports 18 languages, and that audio in other languages falls back to Universal-2 if the request asks for it.
- Amazon's Transcribe product page says its key features work across 100+ languages.
- Deepgram's Nova-3 model language table lists 60 languages, counting the three Chinese variants as one language and German (Switzerland) as a German variant. Deepgram states no total.
- ElevenLabs' speech-to-text documentation says its Scribe v2 model supports 90 or more languages, and lists them with their codes.
- Google Cloud's Speech-to-Text product page says the service supports over 125 languages.
- Microsoft's Azure Speech product page says its audio captioning works in more than 100 languages.
- OpenAI's speech to text guide says Whisper supports 98 languages, and that accuracy varies by language.
- Speechmatics' documentation lists its transcription languages, and counting the table's single-language rows (not the auto option or the seven bilingual and multi-language packs) gives 55 languages for the Enhanced and Standard models.
- Mistral says Voxtral Mini Transcribe V2 transcribes in 13 languages, with speaker diarization, context biasing and word-level timestamps.