无需训练的多步迭代方法，让语音提取更精准
Training-Free Multi-Step Inference for Target Speaker Extraction
- 用插值生成新候选，逐步优化目标说话人语音
- 优化侵入式指标可提升多维度表现，最高增益达1.8dB
- 支持联合优化非侵入指标，适配实际部署需求
目标说话人提取（TSE）旨在利用参考语音从混合语音中恢复目标说话人的声音。现有系统多采用单步推理的条件自编码器架构。受测试时扩展启发，本文提出一种无需训练的多步迭代推理方法，可在冻结预训练模型下实现逐步优化。每一步通过插值原始混合信号与前一估计结果生成新候选，并选取最优者继续迭代直至收敛。实验表明，当有真实目标语音作为监督时，优化侵入式指标（SI-SDRi）能在多个评估指标上持续提升，最高增益达1.8dB；无真值时，优化非侵入指标（UTMOS或SpkSim）仅改善对应指标，可能损害其他表现。为此，我们引入联合指标优化策略，在不同目标间实现平衡，支持实际应用中的可控提取偏好。
原文摘要 · Abstract (English)
Target speaker extraction (TSE) aims to recover a target speaker's speech from a mixture using a reference utterance as a cue. Most TSE systems adopt conditional auto-encoder architectures with one-step inference. Inspired by test-time scaling, we propose a training-free multi-step inference method that enables iterative refinement with a frozen pretrained model. At each step, new candidates are generated by interpolating the original mixture and the previous estimate, and the best candidate is selected for further refinement until convergence. Experiments show that, when ground-truth target speech is available, optimizing an intrusive metric (SI-SDRi) yields consistent gains across multiple evaluation metrics. Without ground truth, optimizing non-intrusive metrics (UTMOS or SpkSim) improves the corresponding metric but may hurt others. We therefore introduce joint metric optimization to balance these objectives, enabling controllable extraction preferences for practical deployment.
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。