Researchers have proposed a new approach to improve the performance of automatic speech recognition (ASR) systems powered by large language models. They suggest sharing connectors between languages that belong to the same linguistic family, rather than training separate connectors for each language. This approach reduces the number of parameters needed and improves generalization across different domains. The study validates its effectiveness using two multilingual language models and real-world speech datasets. --- Why it matters: This matters because it could lead to more efficient and scalable deployment of ASR systems in multilingual environments, which is crucial for applications such as voice assistants and transcription services. Source: https://arxiv.org/abs/2601.18899