Fusing Audio and Metadata Embeddings Improves Language-based Audio Retrieval
This work addresses the challenge of language-based audio retrieval for applications like multimedia search, though it is incremental by building on existing content-based methods.
The paper tackled the problem of matching audio signals with textual descriptions by incorporating audio metadata as an additional clue, resulting in improved retrieval performance with gains of 2.36 and 3.69 percentage points in mAP@10 on ClothoV2 and AudioCaps benchmarks, respectively.
Matching raw audio signals with textual descriptions requires understanding the audio's content and the description's semantics and then drawing connections between the two modalities. This paper investigates a hybrid retrieval system that utilizes audio metadata as an additional clue to understand the content of audio signals before matching them with textual queries. We experimented with metadata often attached to audio recordings, such as keywords and natural-language descriptions, and we investigated late and mid-level fusion strategies to merge audio and metadata. Our hybrid approach with keyword metadata and late fusion improved the retrieval performance over a content-based baseline by 2.36 and 3.69 pp. mAP@10 on the ClothoV2 and AudioCaps benchmarks, respectively.