The Google Cloud Speech-to-Text APITrack this API has been updated to include support for seven new languages, additional enhanced telephony models, broader speech adaptation, and more. The improvements to the API serve to improve the quality and availability of Google’s speech recognition technologies across the globe.
The most noteworthy improvement to the API is added support for seven new languages including Burmese, Estonian, Uzbek, Punjabi, Albanian, Macedonian, and Mongolian. The addition of these languages is especially valuable for applications providing global translation services.
Additionally, Google has expanded support for enhanced telephony models, a method for providing the highest quality transcriptions in situations where audio quality is less than ideal. The company first debuted this option in 2018 and it was originally limited to US English. Google is now releasing support for UK English, Russian, and US Spanish. The announcement of these updates notes that the enhanced model provides 62% better results when compared to the base model.
The Google Cloud Speech-to-Text API includes functionality for speech adaptation, allowing users to customize the pre-built speech models in real-time. This is an important feature helps account for proper nouns and specific product names. Google is releasing the company’s latest iteration of speech adaption, boost based speech adaptation, to 68 new locales.
For additional detail on all the new features included in the Google Cloud Speech-to-Text API make sure to check out Google’s full product update.