MohamedGomaa30/EGYSpeak

EGYSpeak是一个精心整理的埃及阿拉伯语（纯方言）音频数据集，包含147,979个单说话者的音频片段及其转录。音频格式为WAV（PCM_16，16 kHz，单声道），转录语言为埃及阿拉伯语。数据集来源于Kaggle的fadisarwat/egyptian-arabic-lines数据集，并经过严格的ASR流程处理，包括时长过滤、音频处理、说话者分离、ASR转录和质量过滤等步骤。数据集结构包括metadata.csv和metadata.jsonl文件，音频文件被打包成23个tar分片。数据集在CC-BY-4.0许可下发布。

EGYSpeak is a curated dataset of 147,979 single-speaker Egyptian Arabic (pure dialect) audio clips with transcriptions. The audio format is WAV (PCM_16, 16 kHz, mono), and the transcription language is Egyptian Arabic. The dataset is sourced from the fadisarwat/egyptian-arabic-lines Kaggle dataset and processed through a rigorous ASR pipeline, including duration filtering, audio processing, speaker diarization, ASR transcription, and quality filtering. The dataset structure includes metadata.csv and metadata.jsonl files, with audio files packed into 23 tar shards. The dataset is released under the CC-BY-4.0 license.