DOTA-ME-CS
收藏资源简介:
DOTA-ME-CS是由伦敦帝国理工学院、圣安德鲁斯大学、华北电力大学和慕尼黑工业大学合作创建的日常对话文本音频数据集,专注于普通话和英语之间的代码切换。该数据集包含9300条音频记录,总时长为18.54小时,涵盖了34名参与者的日常对话场景。数据集通过人工智能技术(如音色合成、速度变化和噪声添加)增强了多样性和复杂性,旨在为双语语音识别研究提供高质量的资源。数据集的应用领域主要集中在自动语音识别(ASR)系统,特别是解决代码切换带来的挑战,如语言切换点的检测和多语言环境下的语音识别问题。
DOTA-ME-CS is a daily conversational text-audio dataset co-created by Imperial College London, University of St Andrews, North China Electric Power University and Technical University of Munich, focusing on code-switching between Mandarin and English. This dataset contains 9300 audio recordings with a total duration of 18.54 hours, covering daily conversation scenarios involving 34 participants. The diversity and complexity of the dataset are enhanced through artificial intelligence technologies such as timbre synthesis, speed variation and noise addition, aiming to provide high-quality resources for bilingual speech recognition research. Its application domains mainly focus on automatic speech recognition (ASR) systems, particularly addressing challenges brought by code-switching, such as language switch point detection and speech recognition issues in multilingual environments.

- 1DOTA-ME-CS: Daily Oriented Text Audio-Mandarin English-Code Switching Dataset伦敦帝国理工学院, 圣安德鲁斯大学, 华北电力大学, 慕尼黑工业大学 · 2025年



