Zhang, W., Sun, S., Wang, B., Zou, X., Liu, Z., He, Y., Lin, G., Chen, N. F., Aw, A. T. (2025). MoWE-Audio: Multitask AudioLLMs with Mixture of Weak Encoders. ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 1–5. https://doi.org/10.1109/icassp49660.2025.10888128
Abstract:
The rapid advancements in large language models (LLMs) have significantly enhanced natural language processing capabilities, facilitating the development of AudioLLMs that process and understand speech and audio inputs alongside text. Existing AudioLLMs typically combine a pre-trained audio encoder with a pre-trained LLM, which are subsequently finetuned on specific audio tasks. However, the pre-trained audio encoder has constrained capacity to capture features for new tasks and datasets. To address this, we propose to incorporate mixtures of ‘weak’ encoders (MoWE) into the AudioLLM framework. MoWE supplements a base encoder with a pool of relatively lightweight encoders, selectively activated based on the audio input to enhance feature extraction without significantly increasing model size. Our empirical results demonstrate that MoWE effectively improves multi-task performance, broadening the applicability of AudioLLMs to more diverse audio tasks.
License type:
Publisher Copyright
Funding Info:
This research / project is supported by the National Research Foundation, Singapore and Infocomm Media Development Authority, Singapore - National Large Language Models Funding Initiative
Grant Reference no. : NA