Fact · Languages and accents
62languages
In the languages supported by speech recognition systems dataset, the 10 published figures for “Open model” have a median of 62 languages, ranging from 13 to 1,600.
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, 20, 22, 25, 25, 99, 100, 1,107, 1,162, 1,600) = 62.
- 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
- NVIDIA's Canary 1B v2 supports speech transcription for 25 languages, and speech translation from English into 24 languages.
- The IndicVoices authors built IndicASR, which they describe as the first speech recognition model to support all 22 languages in the Eighth Schedule of the Indian Constitution.
- Meta's mms-1b-all checkpoint, on its Hugging Face model card, supports 1162 languages for speech recognition.
- Meta's Massively Multilingual Speech project built a single multilingual speech recognition model for 1,107 languages.
- Meta's Omnilingual ASR is an open-source speech recognition model family for more than 1,600 languages.
- NVIDIA says the parakeet-tdt-0.6b-v3 model extends language support from English to 25 European languages.
- Meta's SeamlessM4T is a single model that supports speech-to-text translation and speech recognition for up to 100 languages.
- The Vosk project says it supports more than 20 languages and dialects, with English, Indian English, German and French among them.
- Mistral says Voxtral Realtime is natively multilingual and transcribes in 13 languages: English, Chinese, Hindi, Spanish, Arabic, French, Portuguese, Russian, German, Japanese, Korean, Italian and Dutch.
- The Whisper paper says its model predicts the spoken language with one token for each language in its training set, and that there are 99 such tokens, so the training data covers 99 languages.