通过联合预测令牌与持续时间实现高效序列转换
Efficient Sequence Transduction by Jointly Predicting Tokens and Durations
AI总结：
本文提出令牌-持续时间转换器（TDT），通过联合预测令牌及其覆盖的输入帧数来跳过冗余帧，在语音识别、翻译和意图分类等任务上同时提升准确率与推理速度，最高加速2.82倍。
摘要：
本文提出了一种新颖的令牌-持续时间转换器（Token-and-Duration Transducer, TDT）架构，用于序列到序列任务。TDT 扩展了传统的 RNN-Transducer 架构，通过联合预测一个令牌及其持续时间，即该输出令牌所覆盖的输入帧数量。这是通过使用一个具有两个输出的联合网络实现的，这两个输出分别独立归一化，以生成令牌分布和持续时间分布。在推理过程中，TDT 模型可以根据预测的持续时间输出跳过输入帧，这使其比逐帧处理编码器输出的传统 Transducer 显著更快。在不同的序列转换任务上，TDT 模型相比传统 Transducer 实现了更高的准确率和显著更快的推理速度。用于语音识别的 TDT 模型相比传统 Transducer 实现了更高的准确率，推理速度最高可提升 2.82 倍。用于语音翻译的 TDT 模型在 MUST-C 测试集上相比传统 Transducer 取得了超过 1 BLEU 的绝对提升，且推理速度提升 2.27 倍。在语音意图分类和槽位填充任务中，TDT 模型相比传统 Transducer 将意图准确率绝对提升最高超过 1%，同时运行速度最高提升 1.28 倍。我们的 TDT 模型实现将随 NeMo（https://github.com/NVIDIA/NeMo）工具包一起开源。
英文摘要：
This paper introduces a novel Token-and-Duration Transducer (TDT) architecture for sequence-to-sequence tasks. TDT extends conventional RNN-Transducer architectures by jointly predicting both a token and its duration, i.e. the number of input frames covered by the emitted token. This is achieved by using a joint network with two outputs which are independently normalized to generate distributions over tokens and durations. During inference, TDT models can skip input frames guided by the predicted duration output, which makes them significantly faster than conventional Transducers which process the encoder output frame by frame. TDT models achieve both better accuracy and significantly faster inference than conventional Transducers on different sequence transduction tasks. TDT models for Speech Recognition achieve better accuracy and up to 2.82X faster inference than conventional Transducers. TDT models for Speech Translation achieve an absolute gain of over 1 BLEU on the MUST-C test compared with conventional Transducers, and its inference is 2.27X faster. In Speech Intent Classification and Slot Filling tasks, TDT models improve the intent accuracy by up to over 1% (absolute) over conventional Transducers, while running up to 1.28X faster. Our implementation of the TDT model will be open-sourced with the NeMo (https://github.com/NVIDIA/NeMo) toolkit.
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。
