fixie-ai/endpointing-audio

该数据集包含四个不同的配置：abstract-tts-podcast-default、common_voice_17_0-en、default和gigaspeech-xl-empty-audio-removed。每个配置都包含音频特征（采样率为16000Hz）、转录文本、对话结束标志（字符串或布尔值）和静音时间（浮点数）。数据集分为训练集和测试集，不同配置的训练集和测试集的大小和示例数量各不相同。

The dataset consists of four different configurations: abstract-tts-podcast-default, common_voice_17_0-en, default, and gigaspeech-xl-empty-audio-removed. Each configuration includes audio features (with a sampling rate of 16000Hz), transcript text, turn completion markers (as string or boolean), and silence after (as float). The dataset is split into training and test sets, with varying sizes and number of examples for the training and test sets across different configurations.