Cascaded Multilingual Audio-Visual Learning from Videos
This work addresses the limitation of prior audio-visual models being limited to English, enabling broader multilingual applications, though it is incremental as it builds on existing self-supervised methods.
The paper tackles the problem of learning multilingual audio-visual representations from videos, proposing a cascaded approach that uses a model trained on English videos to improve performance on other languages like Japanese and Hindi, resulting in a nearly 10x improvement in retrieval performance on Japanese videos and achieving state-of-the-art results on image caption tasks.
In this paper, we explore self-supervised audio-visual models that learn from instructional videos. Prior work has shown that these models can relate spoken words and sounds to visual content after training on a large-scale dataset of videos, but they were only trained and evaluated on videos in English. To learn multilingual audio-visual representations, we propose a cascaded approach that leverages a model trained on English videos and applies it to audio-visual data in other languages, such as Japanese videos. With our cascaded approach, we show an improvement in retrieval performance of nearly 10x compared to training on the Japanese videos solely. We also apply the model trained on English videos to Japanese and Hindi spoken captions of images, achieving state-of-the-art performance.