Towards Automatic Assessment Of Self-supervised Speech Models Using Rank

Abstract

This study explores using embedding rank as an unsupervised evaluation metric for general-purpose speech encoders trained via self-supervised learning (SSL). Traditionally, assessing the performance of these encoders is resource-intensive and requires labeled data from the downstream tasks. Inspired by the vision domain, where embedding rank has shown promise for evaluating image encoders without tuning on labeled downstream data, this work examines its applicability in the speech domain, considering the temporal nature of the signals. The findings indicate rank correlates with downstream performance within encoder layers across various downstream tasks and for in- and out-of-domain scenarios. However, rank does not reliably predict the best-performing layer for specific downstream tasks, as lower-ranked layers can outperform higher-ranked ones. Despite this limitation, the results suggest that embedding rank can be a valuable tool for monitoring training progress in SSL speech models,

Stats

• citations1
• S2 citations—
• github stars0
• HF likes0
• heat score2.26
• arxiv keyaldeneh2024towards

Related papers

• The Efficacy Of Self-supervised Speech Models For Audio Representations (2022)0.00
• Analyzing The Factors Affecting Usefulness Of Self-supervised Pre-trained Representations For Speech Recognition (2022)0.00
• Lebenchmark: A Reproducible Framework For Assessing Self-supervised Representation Learning From Speech (2021)11.39
• Investigation Of Ensemble Features Of Self-supervised Pretrained Models For Automatic Speech Recognition (2022)9.41
• Automatic Pronunciation Assessment Using Self-supervised Speech Representation Learning (2022)0.00
• A Closer Look At Wav2vec2 Embeddings For On-device Single-channel Speech Enhancement (2024)0.00
• Non-contrastive Self-supervised Learning For Utterance-level Information Extraction From Speech (2022)9.59
• Fine-tuning Strategies For Faster Inference Using Speech Self-supervised Models: A Comparative Study (2023)8.35