MSMT-FN: Multi-segment Multi-task Fusion Network for Marketing Audio Classification

TL;DR

This paper introduces MSMT-FN, a novel multi-segment multi-task fusion network that improves audio classification for marketing, demonstrating superior performance on proprietary and benchmark datasets.

Contribution

The paper presents a new multi-segment multi-task fusion network specifically designed for marketing audio classification, with extensive evaluations on proprietary and benchmark datasets.

Findings

MSMT-FN outperforms existing methods on multiple datasets.

The proposed model effectively captures multi-segment audio features.

The MarketCalls dataset is newly curated and available for research.

Abstract

Audio classification plays an essential role in sentiment analysis and emotion recognition, especially for analyzing customer attitudes in marketing phone calls. Efficiently categorizing customer purchasing propensity from large volumes of audio data remains challenging. In this work, we propose a novel Multi-Segment Multi-Task Fusion Network (MSMT-FN) that is uniquely designed for addressing this business demand. Evaluations conducted on our proprietary MarketCalls dataset, as well as established benchmarks (CMU-MOSI, CMU-MOSEI, and MELD), show MSMT-FN consistently outperforms or matches state-of-the-art methods. Additionally, our newly curated MarketCalls dataset will be available upon request, and the code base is made accessible at GitHub Repository MSMT-FN, to facilitate further research and advancements in audio classification domain.