arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.09545cs.SDcs.LGeess.AS
收藏

混响环境下的多麦克风多模态情感识别

Multi-Microphone and Multi-Modal Emotion Recognition in Reverberant Environment

Faculty of Engineering Bar-Ilan University Ramat-Gan, Israel(工程学院 巴伊兰大学 拉马特-甘, 以色列)

Ohad Cohen, Gershon Hazan, Sharon Gannot

AI总结:

针对混响等复杂声学环境下的情感识别难题,提出融合改进HTS-AT多通道音频处理与R(2+1)D视频分析的多模态MER系统,在混响RAVDESS数据集上验证其性能优于单模态及单麦克风方案。

摘要:

本文提出了一种多模态情感识别(Multi-modal Emotion Recognition, MER)系统,旨在提升复杂声学条件下的情感识别准确率。该方法将经改进扩展的层级令牌语义音频Transformer(Hierarchical Token-semantic Audio Transformer, HTS-AT)用于多通道音频处理,同时结合R(2+1)D卷积神经网络(Convolutional Neural Networks, CNN)模型进行视频分析。研究采用合成与真实场景的房间脉冲响应(Room Impulse Responses, RIRs),在瑞尔森情感语音与歌曲视听数据库(Ryerson audio-visual database of emotional speech and song, RAVDESS)的混响版本数据集上评估所提方法。结果表明,与单模态方法相比,融合音频和视频模态可获得更优性能,在复杂声学条件下尤为明显。此外,使用多麦克风的多模态(视听)方法性能优于单麦克风的同类方法。

英文摘要:

This paper presents a Multi-modal Emotion Recognition (MER) system designed to enhance emotion recognition accuracy in challenging acoustic conditions. Our approach combines a modified and extended Hierarchical Token-semantic Audio Transformer (HTS-AT) for multi-channel audio processing with an R(2+1)D Convolutional Neural Networks (CNN) model for video analysis. We evaluate our proposed method on a reverberated version of the Ryerson audio-visual database of emotional speech and song (RAVDESS) dataset using synthetic and real-world Room Impulse Responsess (RIRs). Our results demonstrate that integrating audio and video modalities yields superior performance compared to uni-modal approaches, especially in challenging acoustic conditions. Moreover, we show that the multimodal (audiovisual) approach that utilizes multiple microphones outperforms its single-microphone counterpart.

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。

↑