Parameter-efficient transfer learning (PETL) methods have emerged as a solid alternative to
the standard full fine-tuning approach. They only train a few extra parameters for each …

Audio analysis is useful in many application scenarios. The state-of-the-art audio analysis
approaches assume the data distribution at training and deployment time will be the same …

Abstract Automatic Speech Recognition (ASR) often faces challenges in processing
children's speech due to data scarcity. Training large ASR models becomes particularly …

Abstract Children's Automatic Speech Recognition (ASR) represents a considerable
challenge, with a considerable performance decline of state-of-the-art systems when …

Audio classification plays a crucial role in speech and sound processing tasks with a wide
range of applications. There still remains a challenge of striking the right balance between …

A significant portion of the global population speaks multiple languages, including many low-
resource languages on which current multilingual ASR models perform poorly. To improve …

Despite recent advances in automatic speech recognition (ASR) performance on common
languages, a large fraction of the world's languages remain unsupported. Parameter …