sylviali/EDEN_ASR_Data

EDEN ASR数据集包含来自母语为普通话的说话者的音频片段，这些片段是从与英语练习平台上的聊天机器人对话中收集的。经过过滤后，数据集保留了613次对话中的3081个音频片段，每个片段的长度从1秒到2分钟不等。数据集中的音频片段经过Whisper Medium转录，并经过人工验证以确保转录质量。数据集包含音频URL、情感标签和经过验证的转录文本。数据集的主要用途是用于带口音的自动语音识别（ASR）研究。

The EDEN ASR Dataset contains audio clips of conversations between native Mandarin speakers and a chatbot on an English practice platform. After filtering, the dataset includes 3081 audio clips from 613 conversations and 163 users. The filtering process removed clips containing only Mandarin, duplicates, and a subset of self-introductions. Each audio clip ranges from one second to two minutes. No demographic information was collected to protect user identities. The original audio was transcribed using Whisper Medium, but due to accents, there were instances of ASR errors. These transcripts were manually verified for high quality. The dataset also includes emotion labels, marking some clips as Neutral, Negative, or Pauses. The primary use of the dataset is for accented ASR research.