Zero-Shot Audio Classification using Image Embeddings

TL;DR

This paper explores using image embeddings as semantic side information for zero-shot audio classification, demonstrating that image and textual embeddings can effectively classify unseen audio classes with performance comparable to semantic acoustic embeddings.

Contribution

It introduces a novel approach of leveraging image embeddings for zero-shot audio classification, expanding the types of semantic information used in such models.

Findings

Image embeddings can be used effectively as semantic information for zero-shot audio classification.

Performance of image and textual embeddings is similar and can approach semantic acoustic embeddings.

Semantic similarity between classes significantly influences classification performance.

Abstract

Supervised learning methods can solve the given problem in the presence of a large set of labeled data. However, the acquisition of a dataset covering all the target classes typically requires manual labeling which is expensive and time-consuming. Zero-shot learning models are capable of classifying the unseen concepts by utilizing their semantic information. The present study introduces image embeddings as side information on zero-shot audio classification by using a nonlinear acoustic-semantic projection. We extract the semantic image representations from the Open Images dataset and evaluate the performance of the models on an audio subset of AudioSet using semantic information in different domains; image, audio, and textual. We demonstrate that the image embeddings can be used as semantic information to perform zero-shot audio classification. The experimental results show that the image and textual embeddings display similar performance both individually and together. We additionally calculate the semantic acoustic embeddings from the test samples to provide an upper limit to the performance. The results show that the classification performance is highly sensitive to the semantic relation between test and training classes and textual and image embeddings can reach up to the semantic acoustic embeddings when the seen and unseen classes are semantically similar.