We present a new task, speech dialogue translation mediating speakers of different languages. We construct the SpeechBSD dataset for the task and conduct baseline experiments. Furthermore, we consider context to be an important aspect that needs to be addressed in this task and propose two ways of utilizing context, namely monolingual context and bilingual context. We conduct cascaded speech translation experiments using Whisper and mBART, and show that bilingual context performs better in our settings.

翻译：我们提出一项新任务：面向不同语言使用者的语音对话翻译。我们为这一任务构建了 SpeechBSD 数据集，并开展了基线实验。此外，我们认为上下文是此任务中需要处理的重要方面，并提出了两种利用上下文的方式：单语上下文和双语上下文。我们使用 Whisper 和 mBART 进行了级联语音翻译实验，结果表明，在我们的设置中，双语上下文表现更佳。