arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2102.11474cs.SDeess.AS
收藏

文本到音频的基础:在字幕和声音事件之间建立对应关系

Text-to-Audio Grounding: Building Correspondence Between Captions and Sound Events

Shanghai Jiao Tong University(上海交通大学)

Xuenan Xu, Heinrich Dinkel, Mengyue Wu, Kai Yu

摘要:

自动音频字幕是一个跨模态的任务,生成自然语言描述来总结音频剪辑的声音事件。然而,尚未研究基于其对应的字幕将给定音频中的实际声音事件接地。本文贡献了一个AudioGrounding数据集,它提供了声音事件和Audiocaps中提供的字幕之间的对应关系,以及每个声音事件的位置(时间戳)。在此基础上,我们提出了文本到音频接地(TAG)的任务,交互地考虑音频处理和语言理解之间的关系。提供了基线方法,导致事件F1评分为28.3%,复音检测评分(PSDS)为14.7%。

英文摘要:

Automated Audio Captioning is a cross-modal task, generating natural language descriptions to summarize the audio clips' sound events. However, grounding the actual sound events in the given audio based on its corresponding caption has not been investigated. This paper contributes an AudioGrounding dataset, which provides the correspondence between sound events and the captions provided in Audiocaps, along with the location (timestamps) of each present sound event. Based on such, we propose the text-to-audio grounding (TAG) task, which interactively considers the relationship between audio processing and language understanding. A baseline approach is provided, resulting in an event-F1 score of 28.3% and a Polyphonic Sound Detection Score (PSDS) score of 14.7%.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑