ASMDD: Arabic Speech Mispronunciation Detection Dataset

ASMDD数据集由法尤姆大学和美国开罗大学的研究团队创建，专注于阿拉伯语发音错误的检测。该数据集包含100个埃及儿童（年龄2至8岁）发音的100个最常用阿拉伯语单词的音频文件，总计100条记录。数据集通过Audacity软件录制，具有44.1 kHz的采样率和32位分辨率。创建过程中，孩子们在幼儿园中发音，音频文件随后被分割并标注正确或错误。ASMDD旨在为训练或微调语音表示模型如'wav2vec'和'HuBERT'提供数据，以及促进阿拉伯语发音错误识别技术的发展。

The ASMDD dataset was developed by a research team from Fayoum University and Cairo University (USA), focusing on Arabic pronunciation error detection. This dataset contains audio files of the 100 most commonly used Arabic words pronounced by 100 Egyptian children aged 2 to 8 years old, with a total of 100 records. All audio data was recorded using Audacity software, with a sampling rate of 44.1 kHz and 32-bit resolution. During the dataset creation process, the children pronounced the target words in kindergartens, and the audio files were subsequently segmented and labeled as either correct or incorrect. The ASMDD dataset aims to provide data for training or fine-tuning speech representation models such as wav2vec and HuBERT, and to promote the development of Arabic pronunciation error recognition technologies.