Taiwan Mobile, Changwen Technology Launch Localized myVoca AI Speech Recognition Model

Taiwan Mobile and Changwen Technology jointly developed myVoca, an automatic speech recognition model optimized with Taiwanese language data. Designed for mixed Mandarin, Taiwanese, English and Hakka speech, it aims to narrow recognition gaps involving local accents and code-switching while laying the technical groundwork for voice AI adoption in finance, healthcare and smart government.

The companies recently unveiled myVoca as the first model to support mixed recognition across Mandarin, Taiwanese, English and Hakka. They say it combines low computing requirements with high recognition accuracy and outperforms OpenAI's Whisper-large-v3. They did not disclose a specific launch date, investment amount or quantified accuracy rate, and plan to advance the technology into finance, healthcare and smart government applications.