In this work, we introduce a "score-based assessment" framework for estimating the transferability of pre-trained speech models (PSMs) for fine-tuning target tasks. We leverage upon two representation theories, Bayesian likelihood estimation and optimal transport, to generate rank scores for the PSM candidates using the extracted representations. Our framework efficiently computes transferability scores without actual fine-tuning of candidate models or layers by making a temporal independent hypothesis. We evaluate some popular supervised speech models (e.g., Conformer RNN-Transducer) and self-supervised speech models (e.g., HuBERT) in cross-layer and cross-model settings using public data. Experimental results show a high Spearman's rank correlation and low $p$-value between our estimation framework and fine-tuning ground truth. Our proposed transferability framework requires less computational time and resources, making it a resource-saving and time-efficient approach for tuning speech foundation models.

翻译：在本研究中，我们提出了一种“基于分数的评估”框架，用于估计预训练语音模型（PSMs）在目标任务微调中的迁移能力。我们利用贝叶斯似然估计和最优传输这两种表征理论，通过提取的表征为候选PSM生成排序分数。该框架基于时间独立性假设，无需实际微调候选模型或层即可高效计算迁移性分数。我们使用公开数据在跨层和跨模型设置中评估了若干主流监督语音模型（例如Conformer RNN-Transducer）和自监督语音模型（例如HuBERT）。实验结果表明，我们的评估框架与微调真实值之间具有较高的斯皮尔曼秩相关系数和较低的p值。所提出的迁移性框架所需计算时间和资源较少，因此是一种资源节约且时间高效的语音基础模型调优方法。