nvlachak/aesdd-dataspeech

该数据集是一个专注于语音处理和情感分析的集合，包含605个训练样本，总大小约303KB。数据集特征丰富，涵盖文本内容、说话者身份（如ID和性别）、情感标签（如emotion）、语音声学特征（如音高、音高标准差、语速、音素）、语音质量指标（如PESQ、STOI、SI-SDR）、环境因素（如噪声类型、混响水平）以及语音单调性评估。这些特征支持语音识别、情感检测、语音质量评估和噪声鲁棒性研究，适用于多模态语音分析任务。数据以字符串和数值类型存储，例如文本和分类标签为字符串，声学指标为浮点数。

This dataset is a collection focused on speech processing and emotion analysis, comprising 605 training samples with a total size of approximately 303KB. It features a wide range of attributes, including text content, speaker identity (such as ID and gender), emotion labels, acoustic speech features (e.g., pitch, pitch standard deviation, speaking rate, phonemes), speech quality metrics (e.g., PESQ, STOI, SI-SDR), environmental factors (e.g., noise type, reverberation levels), and speech monotony assessments. These features support tasks like speech recognition, emotion detection, speech quality evaluation, and noise robustness research, making it suitable for multimodal speech analysis. Data is stored in string and numerical types, with text and categorical labels as strings and acoustic metrics as floats.