| 陈欣,张晖.跨文化语境中的多语言语音情感识别[J].,2025,44(4):834-841 |
| 跨文化语境中的多语言语音情感识别 |
| Cross-language speech sentiment recognition based on cultural analysis |
| 投稿时间：2024-08-26 修订日期：2025-06-11 |
| 中文摘要: |
| 为探讨不同文化背景下语言情感表达的差异，并提出一种更加鲁棒的情感识别模型，该文以中国文化和西方文化为研究背景，使用多种情感数据库对多种情感识别器进行了跨语言情感识别实验，并提出了一种双流语声情感识别模型(Dual-TCNet)。该模型通过多层次特征融合，更有效地捕捉了全局和局部信息。实验结果表明，不同语言的情感表达方式存在显著差异，所提出的Dual-TCNet模型在未加权精度和加权精度两个指标上均显著优于基线模型。该模型能够更有效地实现跨语言情感识别，为构建具有广泛适应性的语声情感识别系统提供了新的思路。 |
| 英文摘要: |
| In order to explore the differences in linguistic emotion expression in different cultural contexts and to propose a more robust emotion recognition model, the paper conducted cross-linguistic emotion recognition experiments on multiple emotion recognizers using multiple emotion databases with Chinese and Western cultures as the research background, and proposed a Dual-Stream Speech Acoustic Emotion Recognition Model (Dual-TCNet). The model captures global and local information more effectively through multi-level feature fusion. The experimental results show that there are significant differences in emotion expressions across languages, and the proposed Dual-TCNet model significantly outperforms the baseline model in both unweighted and weighted accuracy metrics. The model is able to realize cross-lingual emotion recognition more effectively, which provides a new idea for constructing a speech emotion recognition system with wide adaptability. |
| DOI：10.11684/j.issn.1000-310X.2025.04.003 |
| 中文关键词: 文化差异 跨语言情感识别 双流模型 语音情感分析 |
| 英文关键词: Cultural differences Cross-language sentiment recognition Dual stream model Sentiment analysis |
| 基金项目: |
|
| 摘要点击次数: 1477 |
| 全文下载次数: 1173 |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
