Wireless Mobile Network with Transfer Learning Algorithm for Multilingual Education and Historical Research
Abstract
Notwithstanding recent advancements in Automatic Speech Recognition (ASR), acknowledging children's speech continues to pose a considerable problem. This mainly results from significant acoustic fluctuation and the scarcity of available data for training in Wireless Mobile Networks (WMN). This issue is especially pronounced in languages other than English, typically under-resourced. This research examines children's ASR in various under-resourced languages by amalgamating different small kids’ voice datasets. Specifically, the study examines the subsequent research questions: Does a novel two-step learning technique comprising Multilingual Learning (MLL) and Historical Research (HR) followed by language-specific Transfer Learning (TL) surpasstraditional single language learning for children's speech and MLL and TL in isolation? Drawing from prior experimental findings with English, the research proposes that MLL enhances the generalization of the fundamental traits of speech in kids. The findings affirmatively address the study issue, demonstrating that TL atop an MLL for an unencountered language surpasses traditional single language-specific instructional techniques.