speechbrain/LargeScaleASR
收藏Hugging Face2025-01-27 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/speechbrain/LargeScaleASR
下载链接
链接失效反馈官方服务：
资源简介：
LargeScaleASR是一个包含25,000小时转录的英语语音识别数据集，旨在为研究和商业用途提供大规模的英语语音识别数据。数据集分为large、medium、small、clean、dev和test六个子集，分别包含不同数量和类型的语音数据。数据来源于多个具有不同许可的现有数据集的混合，包括VoxPopuli、LibriHeavy、Librispeech、YODAS、Peoples Speech和CommonVoice 18.0等。数据集经过文本和音频的标准化处理，确保了较高的数据质量。
LargeScaleASR is an English speech recognition dataset containing 25,000 hours of transcribed speech, designed to provide a large-scale dataset for research and commercial use. The dataset is divided into six subsets: large, medium, small, clean, dev, and test, each containing different amounts and types of speech data. The data is a mixture of multiple existing datasets with different licenses, including VoxPopuli, LibriHeavy, Librispeech, YODAS, Peoples Speech, and CommonVoice 18.0. The dataset has undergone text and audio standardization to ensure high data quality.
提供机构：
speechbrain
