用于端到端 ASR 的说话人平滑 kNN 说话人自适应
Speaker-Smoothed kNN Speaker Adaptation for End-to-End ASR
AI总结：
针对 E2E ASR 中目标说话人数据有限导致的声音失配问题，提出 Speaker-Smoothed kNN，结合数据存储检索与 x-vector 动态插值，在 KeSpeech 和 MagicData 上达到稳定且全域最优的自适应效果。
摘要：
尽管端到端自动语音识别（E2E ASR）系统近期取得了改进，但由于训练数据与测试数据之间的声音特征不匹配，其性能仍可能下降，尤其是在目标说话人自适应数据有限的情况下。我们提出了一种新颖的说话人自适应方法 Speaker-Smoothed kNN，该方法利用 k 近邻（kNN）检索技术，在解码阶段从预先构建的数据存储中查找发音正确的 token，从而改进模型输出。此外，针对数据稀疏问题，我们利用 x-vector 动态调整 kNN 插值参数。该方法在 KeSpeech 和 MagicData 语料库上的域内和全域设置下得到验证。我们的方法始终取得与微调相当的性能，且不会在说话人切换时出现相关的性能下降。此外，在全域设置下，该方法取得了最先进的结果，在单说话人和多说话人测试场景中均降低了 CER。
英文摘要：
Despite recent improvements in End-to-End Automatic Speech Recognition (E2E ASR) systems, the performance can degrade due to vocal characteristic mismatches between training and testing data, particularly with limited target speaker adaptation data. We propose a novel speaker adaptation approach Speaker-Smoothed kNN that leverages k-Nearest Neighbors (kNN) retrieval techniques to improve model output by finding correctly pronounced tokens from its pre-built datastore during the decoding phase. Moreover, we utilize x-vector to dynamically adjust kNN interpolation parameters for data sparsity issue. This approach was validated using KeSpeech and MagicData corpora under in-domain and all-domain settings. Our method consistently performs comparably to fine-tuning without the associated performance degradation during speaker changes. Furthermore, in the all-domain setting, our method achieves state-of-the-art results, reducing the CER in both single speaker and multi-speaker test scenarios.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
