面向流式CTC训练自动语音识别模型的聚焦判别式训练
Focused Discriminative Training For Streaming CTC-Trained Automatic Speech Recognition Models
AI总结:
本文提出 FDT 框架,通过识别并改进音频困难片段的识别,不依赖 HMM 和词格,有效降低了流式 CTC 或 CTC/AED 混合训练的 E2E ASR 模型的 WER。
摘要:
本文提出了一种名为 Focused Discriminative Training (FDT) 的新型训练框架,以进一步改进使用 CTC 或 CTC 与基于注意力的编码器-解码器 (AED) 损失插值训练的流式 word-piece 端到端 (E2E) 自动语音识别 (ASR) 模型。该方法提出了一种新框架,用于识别并改进模型在音频中具有挑战性片段上的识别能力。值得注意的是,该训练框架不依赖于隐马尔可夫模型 (HMMs) 和词格,消除了标准判别式训练方法中通常需要的关于 HMM 拓扑、词典和图生成的重大决策。与在编码器上使用 MMI 或 MWER 损失进行额外微调相比,FDT 在降低于 LibriSpeech 上训练的流式模型的词错误率 (WER) 方面表现出更显著的效果。此外,该方法被证明在进一步改进于 60 万小时助手和听写数据集上训练的已收敛 word-piece 流式 E2E 模型方面同样有效。
英文摘要:
This paper introduces a novel training framework called Focused Discriminative Training (FDT) to further improve streaming word-piece end-to-end (E2E) automatic speech recognition (ASR) models trained using either CTC or an interpolation of CTC and attention-based encoder-decoder (AED) loss. The proposed approach presents a novel framework to identify and improve a model's recognition on challenging segments of an audio. Notably, this training framework is independent of hidden Markov models (HMMs) and lattices, eliminating the need for substantial decision-making regarding HMM topology, lexicon, and graph generation, as typically required in standard discriminative training approaches. Compared to additional fine-tuning with MMI or MWER loss on the encoder, FDT is shown to be more effective in achieving greater reductions in Word Error Rate (WER) on streaming models trained on LibriSpeech. Additionally, this method is shown to be effective in further improving a converged word-piece streaming E2E model trained on 600k hours of assistant and dictation dataset.
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。