Revai/earnings21

Earnings 21数据集是一个包含来自九个不同金融行业公司盈利电话的语料库，总时长为39小时。该语料库旨在为实际环境中的自动语音识别（ASR）系统提供一个基准测试，尤其关注于命名实体识别（NER）。语料库中包含了转码为16KHz的音频文件和格式化文本。README文件详细介绍了数据集的文件格式、实体标签、词错误率（WER）计算方法以及不同系统在该数据集上的结果。此外，README还提供了如何引用该数据集的说明以及相关资源的链接。

The Earnings 21 dataset is a 39-hour corpus of earnings calls from nine different financial sectors, designed to benchmark ASR systems with a focus on NER. It includes transcoded audio files at 16KHz and formatted transcripts. The README details the datasets file formats, entity labels, WER calculation methods, and results from various systems. It also provides guidance on citing the dataset and links to related resources.