突破尖峰：用于加速且精确的自动语音识别的尖峰窗口解码
Breaking Through the Spike: Spike Window Decoding for Accelerated and Precise Automatic Speech Recognition
AI总结：
针对WFST逐帧搜索CTC后验导致推理缓慢的问题，提出Spike Window Decoding，使解码帧数与尖峰帧数线性相关，并在AISHELL-1和大规模内部数据集上达到SOTA准确率且显著加速。
摘要：
近年来，端到端自动语音识别已成为工业界和学术界的主流方法。为优化特定场景下的系统性能，加权有限状态转换器（WFST）被广泛用于整合声学模型和语言模型，利用其在静态图中隐式融合语言模型的能力，从而确保稳健的识别，同时促进快速纠错。然而，WFST需要通过自回归对CTC后验概率进行逐帧搜索，这显著阻碍了推理速度。在本工作中，我们深入研究了CTC输出的尖峰特性，并进一步提出猜想：非空白尖峰的相邻帧携带对模型有益的语义信息。在此基础上，我们提出了Spike Window Decoding（尖峰窗口解码）算法，该算法使WFST中解码的帧数与CTC输出中的尖峰帧数呈线性相关，从而在保证识别性能的同时大幅提高推理速度。我们的方法在AISHELL-1和大规模In-House数据集上均得到验证，实现了SOTA识别准确率并显著加快了解码速度，为将CTC输出与WFST集成建立了一种开创性方法。
英文摘要：
Recently, end-to-end automatic speech recognition has become the mainstream approach in both industry and academia. To optimize system performance in specific scenarios, the Weighted Finite-State Transducer (WFST) is extensively used to integrate acoustic and language models, leveraging its capacity to implicitly fuse language models within static graphs, thereby ensuring robust recognition while also facilitating rapid error correction. However, WFST necessitates a frame-by-frame search of CTC posterior probabilities through autoregression, which significantly hampers inference speed. In this work, we thoroughly investigate the spike property of CTC outputs and further propose the conjecture that adjacent frames to non-blank spikes carry semantic information beneficial to the model. Building on this, we propose the Spike Window Decoding algorithm, which greatly improves the inference speed by making the number of frames decoded in WFST linearly related to the number of spiking frames in the CTC output, while guaranteeing the recognition performance. Our method achieves SOTA recognition accuracy with significantly accelerates decoding speed, proven across both AISHELL-1 and large-scale In-House datasets, establishing a pioneering approach for integrating CTC output with WFST.
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。
