arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2405.02151cs.SDcs.AIeess.AS
收藏

GMP-TL:用于语音情感识别的性别增强多尺度伪标签增强迁移学习

GMP-TL: Gender-augmented Multi-scale Pseudo-label Enhanced Transfer Learning for Speech Emotion Recognition

Kyushu University(九州大学);University of Alberta(阿尔伯塔大学);Ximalaya Inc.(喜马拉雅公司)

Yu Pan, Yuguang Yang, Heng Lu, Lei Ma, Jianjun Zhao

AI总结:

本文提出GMP-TL框架,通过性别增强多尺度伪标签迁移学习结合预训练HuBERT与两阶段微调,在IEMOCAP数据集上WAR达80.0%、UAR达82.0%,超越单模态并媲美多模态SER方法。

摘要:

预训练语音模型的不断演进极大地推动了语音情感识别(SER)。然而,当前研究通常依赖于话语级情感标签,无法充分捕捉单句话语中情感的复杂性。本文提出了GMP-TL,一种新型SER框架,采用基于性别增强多尺度伪标签(GMP)的迁移学习来缓解此问题。具体而言,GMP-TL首先利用预训练HuBERT,通过多任务学习和多尺度k-means聚类获取帧级GMP。随后,为充分利用帧级GMP和话语级情感标签,提出两阶段模型微调方法以进一步优化GMP-TL。在IEMOCAP上的实验表明,GMP-TL的WAR达80.0%,UAR达82.0%,相比最先进的单模态SER方法表现更优,且与多模态SER方法结果相当。

英文摘要:

The continuous evolution of pre-trained speech models has greatly advanced Speech Emotion Recognition (SER). However, current research typically relies on utterance-level emotion labels, inadequately capturing the complexity of emotions within a single utterance. In this paper, we introduce GMP-TL, a novel SER framework that employs gender-augmented multi-scale pseudo-label (GMP) based transfer learning to mitigate this gap. Specifically, GMP-TL initially uses the pre-trained HuBERT, implementing multi-task learning and multi-scale k-means clustering to acquire frame-level GMPs. Subsequently, to fully leverage frame-level GMPs and utterance-level emotion labels, a two-stage model fine-tuning approach is presented to further optimize GMP-TL. Experiments on IEMOCAP show that our GMP-TL attains a WAR of 80.0% and an UAR of 82.0%, achieving superior performance compared to state-of-the-art unimodal SER methods while also yielding comparable results to multimodal SER approaches.

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。

↑