A Large-Scale Evaluation of Speech Foundation Models

TL;DR

This paper introduces SUPERB, a comprehensive benchmark for evaluating speech foundation models, demonstrating their effectiveness across multiple tasks and providing a platform for reproducible, collaborative research in speech processing.

Contribution

It establishes the SUPERB benchmark, proposes a unified multi-tasking framework for speech tasks, and develops a platform for reproducible, community-driven evaluation of speech foundation models.

Findings

Foundation models show promising generalizability across SUPERB tasks.

The multi-tasking framework is simple yet effective for speech tasks.

Benchmarking results are robust and statistically significant.

Abstract

The foundation model paradigm leverages a shared foundation model to achieve state-of-the-art (SOTA) performance for various tasks, requiring minimal downstream-specific modeling and data annotation. This approach has proven crucial in the field of Natural Language Processing (NLP). However, the speech processing community lacks a similar setup to explore the paradigm systematically. In this work, we establish the Speech processing Universal PERformance Benchmark (SUPERB) to study the effectiveness of the paradigm for speech. We propose a unified multi-tasking framework to address speech processing tasks in SUPERB using a frozen foundation model followed by task-specialized, lightweight prediction heads. Combining our results with community submissions, we verify that the foundation model paradigm is promising for speech, and our multi-tasking framework is simple yet effective, as the best-performing foundation model shows competitive generalizability across most SUPERB tasks. For reproducibility and extensibility, we have developed a long-term maintained platform that enables deterministic benchmarking, allows for result sharing via an online leaderboard, and promotes collaboration through a community-driven benchmark database to support new development cycles. Finally, we conduct a series of analyses to offer an in-depth understanding of SUPERB and speech foundation models, including information flows across tasks inside the models, the correctness of the weighted-sum benchmarking protocol and the statistical significance and robustness of the benchmark.