Binaural Speech Synthesis Dataset

该数据集包含2小时的配对单声道和双耳音频数据，由八位不同说话者（四位男性和四位女性）提供，用于M2S转换器的预训练。参与者被要求在一个配备有双耳麦克风的假人周围自然地进行对话。该数据集的规模为2小时音频，其任务是进行音频深度伪造检测。This dataset contains 2 hours of paired monaural and binaural audio data, provided by eight distinct speakers (four male and four female) for pretraining the M2S Transformer. Participants were instructed to conduct natural conversations around a mannequin equipped with binaural microphones. With a total audio duration of 2 hours, this dataset is intended for the task of audio deepfake detection.