数据堂—227小时西班牙语手机采集语音数据_朗读

227小时西班牙语手机采集语音数据由352名来自西班牙、墨西哥、委内瑞拉等国的西班牙语母语人员参与录制；录音环境安静，录音内容丰富，覆盖经济，娱乐，新闻，口语等多个领域。227小时西班牙语手机采集语音数据所有文本均由人工转写，句准确率95%

The 227-hour mobile-collected Spanish speech dataset involves 352 native Spanish speakers recruited from Spain, Mexico, Venezuela and other countries. The recordings were conducted in quiet environments, with rich content covering multiple domains including economy, entertainment, news, spoken language and more. All transcripts of the dataset are manually performed, with a sentence-level accuracy of 95%.

搜集汇总

背景与挑战

背景概述

该数据集包含227小时通过手机采集的西班牙语语音数据，由352名来自西班牙、墨西哥等国的母语者在安静环境下录制，内容涵盖经济、娱乐和新闻等多种主题。数据采用16kHz WAV格式，经过人工标注，句子级准确率达95%，专用于西班牙语语音识别模型的评估任务。