PhoST

PhoST是由VinAI研究创建的高质量大规模英语-越南语语音翻译数据集，包含508小时的音频数据，总计331,284个三元组。数据集内容包括英语源句、越南语目标字幕句及其对应的音频。创建过程涉及音频收集、预处理、句子分割、音频时间戳提取和英越句子对齐等步骤。该数据集旨在推动英语-越南语语音翻译的研究和应用，解决低资源语言翻译问题。

PhoST is a high-quality large-scale English-Vietnamese speech translation dataset created by VinAI Research. It contains 508 hours of audio data and a total of 331,284 triples. The dataset comprises English source sentences, Vietnamese target subtitle sentences and their corresponding audio. Its creation process involves audio collection, preprocessing, sentence segmentation, audio timestamp extraction and English-Vietnamese sentence alignment. This dataset is designed to advance the research and application of English-Vietnamese speech translation, and address the translation challenges for low-resource languages.

搜集汇总

背景与挑战

背景概述

PhoST是一个高质量、大规模的英语-越南语语音翻译数据集，包含508小时音频和33.1万个三元组（音频、英语转录、越南语字幕），适用于机器翻译和语音处理研究，需遵守研究用途限制并引用相关论文。