arXiv 2507.096182025-07-13cs.SDeess.AS

THAI 语音情感识别（THAI-SER）语料库

THAI Speech Emotion Recognition (THAI-SER) corpus

Information Science and Technology（信息科学与技术）；Vidyasirimedhi Institute of Science and Technology（维达亚西姆迪科技研究所）；Chulalongkorn University（朱拉隆功大学）；Department of Dramatic Arts（戏剧艺术系）

Jilamika Wongpithayadisai, Chompakorn Chaksangchaichot, Soravitt Sangnark, Patawee Prakrankamanant, Krit Gangwanpongpun, Siwa Boonpunmongkol, Premmarin Milindas… 展开作者

Jilamika Wongpithayadisai, Chompakorn Chaksangchaichot, Soravitt Sangnark, Patawee Prakrankamanant, Krit Gangwanpongpun, Siwa Boonpunmongkol, Premmarin Milindasuta, Dangkamon Na-Pombejra, Sarana Nutanong, Ekapol Chuangsuwanich

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出首个泰语语音情感识别语料库 THAI-SER，包含 41 小时 36 分钟录音，由 200 名专业演员表演五种情绪，通过众包标注并经严格质量控制，标注者间可靠性达 0.692，人类识别准确率达 0.772，语料库和代码已公开。

摘要：

我们提出了首个规模可观的泰语语音情感识别语料库 THAI-SER，包含来自 100 次录音的 41 小时 36 分钟（27,854 条话语），这些录音在不同录制环境中进行：Zoom 和两种录音棚设置。录音包含脚本和即兴表演环节，由 200 名专业演员（112 名女性和 88 名男性，年龄在 18 至 55 岁之间）表演，并由专业导演指导。有五种主要情绪：中性、愤怒、高兴、悲伤和沮丧，在录制话语时分配给演员。话语通过众包进行情感类别标注。为了控制标注过程的质量，我们还设计了一套广泛的过滤和质量控制方案，以确保多数同意分数保持在 0.71 以上。我们使用两个指标评估标注语料库：标注者间可靠性和人类识别准确率。标注者间可靠性分数使用 Krippendorff's alpha 计算，我们的语料库在过滤后达到 0.692 的 alpha 分数，高于 0.667 的推荐值。对于人类识别准确率，我们的语料库在过滤后得分高达 0.772。我们还提供了在该语料库上训练的模型在语料库内和跨语料库设置下评估的结果。该语料库在 Creative Commons BY-SA 4.0 许可下公开提供，以及我们的实验代码。

英文摘要：

We present the first sizeable corpus of Thai speech emotion recognition, THAI-SER, containing 41 hours and 36 minutes (27,854 utterances) from 100 recordings made in different recording environments: Zoom and two studio setups. The recordings contain both scripted and improvised sessions, acted by 200 professional actors (112 females and 88 males, aged 18 to 55) and were directed by professional directors. There are five primary emotions: neutral, angry, happy, sad, and frustrated, assigned to the actors when recording utterances. The utterances are annotated with an emotional category using crowdsourcing. To control the annotation process's quality, we also design an extensive filtering and quality control scheme to ensure that the majority agreement score remains above 0.71. We evaluate our annotated corpus using two metrics: inter-annotator reliability and human recognition accuracy. Inter-annotator reliability score was calculated using Krippendorff's alpha, where our corpus, after filtering, achieved an alpha score of 0.692, higher than a recommendation of 0.667. For human recognition accuracy, our corpus scored up to 0.772 post-filtering. We also provide the results of the model trained on the corpus evaluated on both in-corpus and cross-corpus setups. The corpus is publicly available under a Creative Commons BY-SA 4.0, as well as our codes for the experiments.

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。