Pansori-TEDxKR

Pansori-TEDxKR是由仁荷大学开发的韩语自动语音识别（ASR）数据集，基于韩国TED会议演讲和社区转录的字幕。该数据集包含3091条高质量的音频文本对，旨在为独立研究提供首个免费的高质量韩语ASR资源。数据集的创建过程涉及从在线视频内容中提取音频和字幕，通过云端语音API进行对齐和验证，确保数据准确性。Pansori-TEDxKR的应用领域广泛，主要用于提高韩语ASR模型的准确性和性能，解决韩语ASR研究中数据稀缺的问题。

Pansori-TEDxKR is a Korean automatic speech recognition (ASR) dataset developed by Inha University, based on Korean TED conference speeches and community-transcribed subtitles. This dataset contains 3,091 high-quality audio-text pairs, aiming to provide the first free, high-quality Korean ASR resource for independent research. The dataset creation process involves extracting audio and subtitles from online video content, and conducting alignment and verification via cloud speech APIs to ensure data accuracy. Pansori-TEDxKR has a wide range of application scenarios, primarily used to enhance the accuracy and performance of Korean ASR models and address the issue of data scarcity in Korean ASR research.

搜集汇总

背景与挑战

背景概述

Pansori-TEDxKR是由仁荷大学开发的韩语自动语音识别数据集，包含3091条基于TED演讲的高质量音频文本对，旨在提供首个免费韩语ASR资源以解决数据稀缺问题。其创建过程通过音频提取和对齐验证确保准确性，主要应用于提升韩语ASR模型的性能。