SpeechEQ数据集(SEQD)
收藏资源简介:
SpeechEQ数据集(SEQD)是由平安科技(深圳)有限公司创建的一个内容无关的普通话语音情感识别数据集。该数据集包含总计2.3小时的语音,来自20位说话者(10男10女)的1648个音频片段,采样率为16kHz,精度为16位。数据集的创建过程涉及每位说话者独立编写描述情感的句子,并通过三位独立评审的验证确保情感表达的准确性。SEQD数据集旨在通过提供统一的情感识别标准,改善智能系统对人类情感状态的理解和交互能力,主要应用于语音情感识别领域。
SpeechEQ Dataset (SEQD) is a content-independent Mandarin speech emotion recognition dataset created by Ping An Technology (Shenzhen) Co., Ltd. It contains a total of 2.3 hours of speech, including 1648 audio clips from 20 speakers (10 males and 10 females), with a sampling rate of 16 kHz and 16-bit precision. The dataset creation process involved each speaker independently writing sentences that describe emotions, and the accuracy of the expressed emotions was verified by three independent reviewers. The SEQD dataset aims to provide a unified standard for emotion recognition to improve intelligent systems' understanding of human emotional states and their interaction capabilities, and is mainly applied in the field of speech emotion recognition.




