一个模型，多种延迟：面向多样化实时应用的通用语音增强

摘要

不同的实时语音应用对延迟预算有着严格的要求，通常需要为每种场景单独训练增强模型。本文提出了一种实时通用语音增强模型，能够同时显式控制算法延迟和计算延迟。通过可配置的前瞻帧数灵活调整算法延迟，为避免因不同填充配置导致的学习效率低下问题，我们引入了对应不同前瞻设置的并行卷积层。计算延迟通过早停机制实现控制，允许在不同网络深度进行推理。为缩小专用模型与灵活模型之间的性能差距，我们提出了一种包含共享到多解码器过渡的两阶段训练策略。总体而言，该框架使得单个模型能够在不同延迟预算下部署，无需重新训练独立模型。

English

Different real-time speech applications impose distinct latency budgets, often requiring separately trained enhancement models for each scenario. In this paper, we propose a one-for-all, real-time universal speech enhancement model that provides explicit control over both algorithmic and computational latency. Algorithmic latency is flexibly adjusted via configurable look-ahead frames. To avoid learning inefficiency caused by varying padding configurations, we introduce parallel convolutional layers corresponding to different look-ahead settings. Computational latency is controlled through an early-exit mechanism, enabling inference at different network depths. To narrow the performance gap between specialized and flexible models, we propose a two-stage training strategy with a shared-to-multiple decoder transition. Overall, the proposed framework enables a single model to be deployed across diverse latency budgets without retraining separate models.