yigagilbert/synthetic-parallel-v4

这是一个双语语音数据集，包含英语（audio_eng, text_eng）和卢干达语（audio_lug, text_lug）的音频文件及其对应的文本转录。数据集还包括每个样本的唯一ID、说话者信息（speaker_eng, speaker_lug）、源音频和目标音频的时长（src_dur_s, tgt_dur_s）、时长比率（dur_ratio）、语音比率（src_speech_ratio, tgt_speech_ratio），以及来源数据集和子集信息（source_dataset, source_subset）。数据集分为训练集（181,401个示例）和验证集（9,548个示例），总大小约为109.35 GB，下载大小约为107.87 GB。可能用于语音识别、机器翻译或跨语言语音处理任务。This is a bilingual speech dataset containing audio files and corresponding text transcriptions in English (audio_eng, text_eng) and Luganda (audio_lug, text_lug). The dataset also includes unique IDs for each sample, speaker information (speaker_eng, speaker_lug), durations of source and target audio (src_dur_s, tgt_dur_s), duration ratio (dur_ratio), speech ratios (src_speech_ratio, tgt_speech_ratio), and source dataset and subset information (source_dataset, source_subset). It is split into a training set (181,401 examples) and a validation set (9,548 examples), with a total size of approximately 109.35 GB and a download size of approximately 107.87 GB. It is likely intended for tasks such as speech recognition, machine translation, or cross-lingual speech processing.