Arabic Natural Audio Dataset

This is the first Arabic Natural Audio Dataset (ANAD) developed to recognize 3 discrete emotions: Happy,angry, and surprised.

Eight videos of live calls between an anchor and a human outside the studio were downloaded from online Arabic talk shows. Each video was then divided into turns: callers and receivers. To label each video, 18 listeners were asked to listen to each video and select whether they perceive a happy, angry or surprised emotion. Silence, laughs and noisy chunks were removed. Every chunk was then automatically divided into 1 sec speech units forming our final corpus composed of 1384 records.

Twenty five acoustic features, also known as low-level descriptors, were extracted. These features are: intensity, zero crossing rates, MFCC 1-12 (Mel-frequency cepstral coefficients), F0 (Fundamental frequency) and F0 envelope, probability of voicing and, LSP frequency 0-7. On every feature nineteen statistical functions were applied. The functions are: maximum, minimum, range, absolute position of maximum, absolute position of minimum, arithmetic of mean, Linear Regression1, Linear Regression2, Linear RegressionA, Linear RegressionQ, standard Deviation, kurtosis, skewness, quartiles 1, 2, 3 and, inter-quartile ranges 1-2, 2-3, 1-3. The delta coefficient for every LLD is also computed as an estimate of the first derivative hence leading to a total of 950 features.

本数据集为首个阿拉伯语自然音频数据集（ANAD），旨在识别三种离散情绪：快乐、愤怒和惊讶。该数据集由在线阿拉伯谈话节目中的八段主播与场外人士的实时通话视频组成。每段视频随后被细分为通话者与接收者两个部分。为标注每段视频，我们邀请了18名听众聆听并选择他们感知到的情绪类型——快乐、愤怒或惊讶。视频中出现的静默、笑声和嘈杂片段均被移除。每个片段随后被自动分割为1秒的语音单元，最终形成包含1384条记录的语料库。

数据集中提取了25个声学特征，亦称为低级描述符，包括：强度、过零率、MFCC 1-12（梅尔频率倒谱系数）、基频及其包络、发音概率以及LSP频率0-7。针对每个特征，应用了19个统计函数，包括：最大值、最小值、范围、最大值的绝对位置、最小值的绝对位置、算术平均值、线性回归1、线性回归2、线性回归A、线性回归Q、标准差、峰度、偏度、四分位数1、2、3以及四分位距1-2、2-3、1-3。每个低级声学特征（LLD）的增量系数也被计算，作为一阶导数的估计，从而总计形成950个特征。