用最小贝叶斯风险提升非自回归语音识别速度与准确率
Non-Autoregressive Minimum Bayes' Risk Decoding for Fast Speech Recognition
- 基于最小贝叶斯风险，从模型输出中采样并优化期望收益
- 在LibriSpeech等数据集上超越已有非自回归方法，且快于自回归解码
- 适合追求高速高精度语音识别的工业应用
非自回归（NAR）解码通过并行生成输出标记，使语音识别速度优于自回归（AR）解码。然而，由于无法依赖已生成标记来消除不确定性，识别性能下降。为此，我们提出一种基于最小贝叶斯风险（MBR）的新型NAR解码框架——NAR-MBR解码，其通过从NAR模型输出概率中采样，最大化预期效用而非输出概率本身。值得注意的是，借助NAR模型特性，可在一次前向计算中高效获得多个样本。在LibriSpeech、Switchboard、AMI及网络演讲语料库上的实验表明，该方法优于现有NAR解码，且运行速度超过AR解码。
原文摘要 · Abstract (English)
Non-autoregressive (NAR) decoding generates output tokens in parallel, making speech recognition faster than autoregressive decoding, which generates them sequentially from left to right. However, the recognition performance is degraded because NAR decoding cannot resolve uncertainty by conditioning on previously generated tokens. To address this issue, we propose a novel NAR decoding framework based on minimum Bayes' risk (MBR) decoding, termed NAR-MBR decoding, that maximizes the expected utility calculated from samples drawn from the output probability of an NAR model rather than maximizing the output probability. Notably, by leveraging the nature of NAR models, multiple samples are obtained efficiently with a single forward computation. Our experiments across LibriSpeech, Switchboard, AMI, and web presentation corpus demonstrated that our NAR-MBR decoding outperformed previous NAR decoding and ran faster than AR decoding.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。