Google has added voice recognition for 30 new languages, which it says opens voice-based services to more than a billion people.
With the additional languages, Google speech recognition supports 119 language varieties in features like its virtual keyboard Gboard on Android and in voice search.
In a blog post, Google said it worked with native speakers to collect speech samples and read common phrases of the new languages.
“This process trained our machine learning models to understand the sounds and words of the new languages and to improve their accuracy when exposed to more examples over time,” wrote Daan van Esch, technical program manager of speech at Google, in the post.
The new languages and locales include:
- Amharic (Ethiopia)
- Armenian (Armenia)
- Azerbaijani (Azerbaijani)
- Bengali (Bangladesh, India)
- English (Ghana, Kenya, Nigeria, Tanzania)
- Georgian (Georgia)
- Gujarati (India)
- Javanese (Indonesia)
- Kannada (India)
- Khmer (Cambodian)
- Lao (Laos)
- Latvian (Latvia)
- Malayalam (India)
- Marathi (India)
- Nepali (Nepal)
- Sinhala (Sri Lanka)
- Sundanese (Indonesia)
- Swahili (Tanzania, Kenya)
- Tamil (India, Singapore, Sri Lanka, Malaysia)
- Telugu (India)
- Urdu (Pakistan, India)
The new languages are also available in Google’s Cloud Speech API for developers and will soon be available for other Google products, like its Translate app.
“To honor languages around the world, speech recognition will support ancient languages such as Georgian, which has an alphabet that dates back to the 10th century,” van Esch wrote. “We’re also adding Swahili and Amharic, two of Africa's largest languages, as well as many Indian languages, on our quest to make the internet more inclusive.”
In addition, Google said English speakers in the US can now use voice dictation to express themselves with emoji by simply speaking the name of the emoji they’d like to use, like, “winky face emoji.” Google said this functionality will be available for more languages soon.
