Mandarin English Code Switching

Mandarin-English code-switching research focuses on improving automatic speech recognition (ASR) systems' ability to accurately transcribe speech containing spontaneous switches between Mandarin and English. Current efforts concentrate on adapting existing multilingual models like Whisper, employing techniques such as mixture-of-experts architectures and attention mechanisms to better handle language boundaries and context. These advancements are crucial for creating more robust and accurate ASR systems for real-world applications, particularly in multilingual communities where code-switching is prevalent, and for advancing our understanding of how humans process mixed-language speech.