An Automatic Speech Recognition Model for Turkic Languages was Presented in Kazakhstan
An Automatic Speech Recognition Model for Turkic Languages was Presented in Kazakhstan
19 January
The Kazakh startup Cybernet AI released the largest automatic speech recognition (ASR) model for Turkic languages. It is the first example of a fully-fledged AI model originally created for a specific language group, rather than adapted from English-language solutions. Information about this was posted on the Unified Platform of Internet Resources of Government Agencies of Kazakhstan.
The model was developed using the Kazakh, Tatar, Turkish, Uzbek, Kyrgyz, and Azerbaijani languages. It accurately recognizes mixed Turkic-Russian speech, which is common in real-life customer conversations with businesses and when receiving government services.
The technology is designed for companies that handle large volumes of voice communications: banks and financial institutions, telecoms, contact centers, service companies, and government agencies.
The model is resistant to noise, accents and switching languages within a single sentence.
The model was developed using the Kazakh, Tatar, Turkish, Uzbek, Kyrgyz, and Azerbaijani languages. It accurately recognizes mixed Turkic-Russian speech, which is common in real-life customer conversations with businesses and when receiving government services.
The technology is designed for companies that handle large volumes of voice communications: banks and financial institutions, telecoms, contact centers, service companies, and government agencies.
The model is resistant to noise, accents and switching languages within a single sentence.
