swdq/asr_correct

该数据集主要用于自动语音识别（ASR）任务，特别是针对日语内容，并且包含NSFW（不适合工作场所）和视觉相关的标签。数据集的大小在10万到100万之间。数据集由多个子数据集组成，包括OOPPEENN/Galgame_Dataset、grider-withourai/nekopara-speech和litagin/Galgame_Speech_ASR_16kHz。数据集的目的是通过使用whisper模型的转录结果和正确标签来解决whisper在NSFW内容转录中的不准确性。

This dataset is primarily used for automatic speech recognition tasks, especially for NSFW (Not Safe For Work) content and visual content. The dataset includes multiple sub-datasets such as OOPPEENN/Galgame_Dataset, grider-withourai/nekopara-speech, and litagin/Galgame_Speech_ASR_16kHz. The purpose of these datasets is to improve the inaccuracies of the Whisper model in transcribing NSFW content by using the Whisper model for speech-to-text and combining it with correct labels. The size of the dataset is between 100K and 1M, and the language is Japanese.