WavLabLM Multilingual Data

该数据集包含了40,000小时的多语种数据，用于对WavLabLM模型进行预训练。该数据集解决了语言不平衡的问题，并支持包括自动语音识别和语言识别在内的多种多语种任务。其规模达到了40,000小时，专用于多语种的自监督学习任务。This dataset comprises 40,000 hours of multilingual data, which is utilized for pre-training the WavLabLM model. It addresses the issue of language imbalance and supports a variety of multilingual tasks including automatic speech recognition and language identification. With a total duration of 40,000 hours, this dataset is specifically designed for multilingual self-supervised learning tasks.