Audio-Visual Approach For Multimodal Concurrent Speaker Detection

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Eliav, Amit, Gannot, Sharon
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866910784849707008
author Eliav, Amit
Gannot, Sharon
author_facet Eliav, Amit
Gannot, Sharon
contents Concurrent Speaker Detection (CSD), the task of identifying active speakers and their overlaps in an audio signal, is essential for various audio applications, including meeting transcription, speaker diarization, and speech separation. This study presents a multimodal deep learning approach that integrates audio and visual information. The proposed model utilizes an early fusion strategy, combining audio and visual features through cross-modal attention mechanisms with a learnable [CLS] token to capture key audio-visual relationships. The model is extensively evaluated on two real-world datasets, the established AMI dataset and the recently introduced EasyCom dataset. Experiments validate the effectiveness of the multimodal fusion strategy. An ablation study further supports the design choices and the model's training procedure. As this is the first work reporting CSD results on the challenging EasyCom dataset, the findings demonstrate the potential of the proposed multimodal approach for \ac{CSD} in real-world scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2407_01774
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Audio-Visual Approach For Multimodal Concurrent Speaker Detection
Eliav, Amit
Gannot, Sharon
Audio and Speech Processing
Image and Video Processing
Concurrent Speaker Detection (CSD), the task of identifying active speakers and their overlaps in an audio signal, is essential for various audio applications, including meeting transcription, speaker diarization, and speech separation. This study presents a multimodal deep learning approach that integrates audio and visual information. The proposed model utilizes an early fusion strategy, combining audio and visual features through cross-modal attention mechanisms with a learnable [CLS] token to capture key audio-visual relationships. The model is extensively evaluated on two real-world datasets, the established AMI dataset and the recently introduced EasyCom dataset. Experiments validate the effectiveness of the multimodal fusion strategy. An ablation study further supports the design choices and the model's training procedure. As this is the first work reporting CSD results on the challenging EasyCom dataset, the findings demonstrate the potential of the proposed multimodal approach for \ac{CSD} in real-world scenarios.
title Audio-Visual Approach For Multimodal Concurrent Speaker Detection
topic Audio and Speech Processing
Image and Video Processing
url https://arxiv.org/abs/2407.01774