Taiwanese Hakka Across Taiwan Corpus and Formosa Speech Recognition Challenge 2023 - Hakka ASR.

To revive the endangered Taiwanese Hakka language, the first large-scale Taiwanese Hakka speech corpus across Taiwan (HAT) was developed, representing modern Taiwanese Hakka around Taiwan. This paper briefly reviews the first part (Sixian and Hailu accents) of the HAT corpus and demonstrates some of its potential applications, including automatic speech recognition (ASR), text-to-speech (TTS), and Taiwanese Hakka speech-enabled ChatGPT. Moreover, the Formosa Speech Recognition Challenge 2023 (FSR-2023) - Hakka ASR was established to promote the corpus and evaluate the performance of state-of-the-art Taiwanese Hakka ASR systems.