A Light-Weight Multimodal Framework for Improved Environmental Audio Tagging

TL;DR

This paper introduces a lightweight multimodal framework combining audio and visual data for environmental audio tagging, achieving improved accuracy with efficient training on large weakly labeled datasets.

Contribution

It presents a novel, resource-efficient multimodal system that integrates pretrained visual models and weakly labeled audio data for enhanced environmental sound tagging.

Findings

Video information improves F1 score by 5.3%.

System trains within 6 hours on a single GPU.

Framework adaptable to other audio tasks.

Abstract

The lack of strong labels has severely limited the state-of-the-art fully supervised audio tagging systems to be scaled to larger dataset. Meanwhile, audio-visual learning models based on unlabeled videos have been successfully applied to audio tagging, but they are inevitably resource hungry and require a long time to train. In this work, we propose a light-weight, multimodal framework for environmental audio tagging. The audio branch of the framework is a convolutional and recurrent neural network (CRNN) based on multiple instance learning (MIL). It is trained with the audio tracks of a large collection of weakly labeled YouTube video excerpts; the video branch uses pretrained state-of-the-art image recognition networks and word embeddings to extract information from the video track and to map visual objects to sound events. Experiments on the audio tagging task of the DCASE 2017 challenge show that the incorporation of video information improves a strong baseline audio tagging system by 5.3\% absolute in terms of $F_1$ score. The entire system can be trained within 6~hours on a single GPU, and can be easily carried over to other audio tasks such as speech sentimental analysis.