https://scholars.lib.ntu.edu.tw/handle/123456789/607151
Title: | SUPERB: Speech processing Universal PERformance Benchmark | Authors: | Yang S.-W HUNG-YI LEE et al. |
Keywords: | Benchmark;Evaluation;Model generalization;Representation learning;Self-supervised learning;Speech;Benchmarking;Learning algorithms;Natural language processing systems;Speech communication;Supervised learning;Large volumes;Performance;Shared model;State of the art;Unlabeled data;Speech processing | Issue Date: | 2021 | Journal Volume: | 4 | Start page/Pages: | 3161-3165 | Source: | Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH | Abstract: | Self-supervised learning (SSL) has proven vital for advancing research in natural language processing (NLP) and computer vision (CV). The paradigm pretrains a shared model on large volumes of unlabeled data and achieves state-of-the-art (SOTA) for various tasks with minimal adaptation. However, the speech processing community lacks a similar setup to systematically explore the paradigm. To bridge this gap, we introduce Speech processing Universal PERformance Benchmark (SUPERB). SUPERB is a leaderboard to benchmark the performance of a shared model across a wide range of speech processing tasks with minimal architecture changes and labeled data. Among multiple usages of the shared model, we especially focus on extracting the representation learned from SSL for its preferable re-usability. We present a simple framework to solve SUPERB tasks by learning task-specialized lightweight prediction heads on top of the frozen shared model. Our results demonstrate that the framework is promising as SSL representations show competitive generalizability and accessibility across SUPERB tasks. We release SUPERB as a challenge with a leaderboard1 and a benchmark toolkit2 to fuel the research in representation learning and general speech processing. Copyright ? 2021 ISCA. |
URI: | https://www.scopus.com/inward/record.uri?eid=2-s2.0-85119172130&doi=10.21437%2fInterspeech.2021-1775&partnerID=40&md5=c0544332855c4f5d60a2d48455cc8099 https://scholars.lib.ntu.edu.tw/handle/123456789/607151 |
ISSN: | 2308457X | DOI: | 10.21437/Interspeech.2021-1775 |
Appears in Collections: | 電機工程學系 |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.