Sinoosoida/speech_ru

这是一个包含俄语有声读物的数据集，其中每个有声读物都经过语音识别处理，并将识别结果作为数据集的一部分。数据集将每个有声读物分割成多个片段，并建立了这些片段与书籍文本版本相应片段的对应关系。

This dataset consists of Russian-language audiobooks that have been scraped from the internet, with speech-to-text (STT) results included for each audiobook. Each audiobook is divided into segments, and a correspondence between the transcribed text and the original book text is established using word-level Levenshtein distance, matching each audio segment with the corresponding fragment in the text version of the book.