Omni-Router: Sharing Routing Decisions in Sparse Mixture-of-Experts for Speech Recognition

TL;DR

The paper introduces Omni-router, a shared routing mechanism for sparse MoE models in speech recognition, leading to better expert cooperation, lower error rates, and enhanced robustness across diverse datasets.

Contribution

It proposes a shared router across MoE layers to improve expert collaboration and specialization in speech recognition models.

Findings

Achieves 11.2% reduction in word error rate compared to dense models.

Outperforms Switch Transformer with 8.2% lower word error rate.

Demonstrates improved robustness across 10 out-of-domain benchmarks.

Abstract

Mixture-of-experts (MoE) architectures have expanded from language modeling to automatic speech recognition (ASR). Traditional MoE methods, such as the Switch Transformer, route experts independently within each layer. Our analysis reveals that routers in most layers make expert choices that are not strongly correlated with the choices of the routers in other layers. To increase the cooperation between experts in different layers and encourage greater specialization, we use a shared router across different MoE layers. We call this model Omni-router Transformer. Extensive experiments on a large-scale pseudo-labeled dataset and evaluations across 10 diverse, out-of-domain ASR benchmarks demonstrate that the Omni-router Transformer is able to achieve lower training loss and consistently outperform dense and Switch Transformer models, reducing average word error rates by 11.2% and 8.2%, respectively, while providing structured expert usage and improved robustness to diverse data.