ntnu-smil/unseen_1964_qa

该数据集包含了语音相关特征，如说话人ID、音频路径、语音转文字、各种分数（如信心分数、声学分数、语言模型分数等）、语音的基频、强度、停顿时间、静默时间、持续时间、每分钟单词数、单词总数、语音段长度、关键字分数、平均值、阈值计数、平均基频、平均强度、持续时间、局部抖动、局部闪烁、快速抖动、长静默、静默、长静默数、静默数、能量标准差、平均频谱、平均能量熵、零交叉数、清音与浊音比例、语音帧数、非语音帧数、平均长静默、平均静默、三个以上单词数量、单词数、 WhisperX转录、发音向量、句子数、uh词数量、静默数量、长静默数量、语音识别结果、四个响应及其分数、四个相似度分数等。数据集分为训练集、验证集、测试集和完整测试集，分别包含不同数量的示例。

The dataset includes voice-related features such as speaker ID, audio path, speech-to-text transcription, various scores (such as confidence score, acoustic score, language model score, etc.), voice pitch, intensity, pause time, silence time, duration, words per minute, total number of words, voice segment length, key scores, average value, threshold count, average pitch, average intensity, duration, local jitter, local shimmer, rapid jitter, long silence, silence, long silence number, silence number, standard deviation of energy, average spectrum, average energy entropy, zero crossing number, voicing to unvoicing ratio, voice frame count, unvoice frame count, average long silence, average silence, number of words with more than three characters, total number of words, WhisperX transcription, delivery vector, number of sentences, uh word count, number of silences, number of long silences, automatic speech recognition results, four responses and their scores, four similarity scores, etc. The dataset is split into training set, validation set, test set, and full test set, each containing a different number of examples.