arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2310.04791eess.AScs.LGcs.SD
收藏

面向目标说话人提取的条件扩散模型

Conditional Diffusion Model for Target Speaker Extraction

University of Cambridge Research Computing Service(剑桥大学研究计算服务中心)

Theodor Nguyen, Guangzhi Sun, Xianrui Zheng, Chao Zhang, Philip C Woodland

AI总结:

本研究提出基于条件扩散模型的目标说话人提取方法DiffSpEx,在复短时傅里叶变换域建模扩散过程,结合说话人嵌入提取目标语音,在WSJ0-2mix数据集取得良好效果,微调可实现个性化提取。

摘要:

我们提出DiffSpEx,这是一种基于随机微分方程的基于分数的生成建模的生成式目标说话人提取方法。DiffSpEx在复短时傅里叶变换域中部署连续时间随机扩散过程,从目标说话人源开始,收敛到以源混合信号为中心的高斯分布。对于逆时过程,参数化分数函数以目标说话人嵌入为条件,从源混合信号中提取目标说话人。我们使用ECAPA-TDNN目标说话人嵌入,并交替以SDE时间嵌入和目标说话人嵌入作为分数函数的条件。我们在WSJ0-2mix数据集上验证了DiffSpEx的潜力,取得了12.9 dB的SI-SDR和3.56的NISQA分数。此外,我们表明将预训练的DiffSpEx模型微调到特定说话人可进一步提升性能,实现目标说话人提取的个性化。

英文摘要:

We propose DiffSpEx, a generative target speaker extraction method based on score-based generative modelling through stochastic differential equations. DiffSpEx deploys a continuous-time stochastic diffusion process in the complex short-time Fourier transform domain, starting from the target speaker source and converging to a Gaussian distribution centred on the mixture of sources. For the reverse-time process, a parametrised score function is conditioned on a target speaker embedding to extract the target speaker from the mixture of sources. We utilise ECAPA-TDNN target speaker embeddings and condition the score function alternately on the SDE time embedding and the target speaker embedding. The potential of DiffSpEx is demonstrated with the WSJ0-2mix dataset, achieving an SI-SDR of 12.9 dB and a NISQA score of 3.56. Moreover, we show that fine-tuning a pre-trained DiffSpEx model to a specific speaker further improves performance, enabling personalisation in target speaker extraction.

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。

↑