• KATIKIDALA GOWTHAMI SRI
• Dr. K. Jhansi Rani

Emotions unfold as continuous streams rather than isolated snapshots, and research into detecting the temporal boundaries of emotion segments in audio-visual signals has produced multi-stage pipelines involving boundary proposal, classification, completeness checking, and overlap suppression (Li et al., 2024). Reproducing such pipelines end to end typically requires a trained multimodal model and a curated dataset, which may place hands-on exploration of the workflow beyond the practical reach of most learners and early-stage developers. This report documents AVES Pulse, a native Android application built with Kotlin and Jetpack Compose that reconstructs the shape of this workflow as an interactive, fully local mobile tool. The application enables users to capture or select a video, observe its progression through a five-stage simulated detection pipeline, and inspect the resulting emotion segments on an interactive timeline, run multi-annotator agreement analysis using Fleiss' Kappa, and benchmark configurations using temporal Intersection-over-Union (tIoU) and Average Precision (AP) / mean Average Precision (mAP). The application follows a Model–View–ViewModel (MVVM) architecture with Room for local persistence, CameraX and AudioRecord for live audio-visual capture, and an optional cloud language-model call as one data source among several. Verification against the app's own bundled demo data suggests that the statistical modules — Fleiss' Kappa, tIoU, and AP/mAP — appear to be computed correctly, while the detection stages themselves appear to remain a UI-level simulation intended for demonstration, teaching, and rapid prototyping rather than to constitute a trained model. The report describes the system architecture, the pipeline design, the statistics implemented, and a roadmap toward replacing the simulated stages with genuine on-device inference.

Android, Jetpack Compose, MVVM Architecture, Emotion Stream, Fleiss' Kappa, Temporal IoU, Average Precision, Room Database, Mobile Prototyping

"Audio Visual Emotion Detection System", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 10, page no.a192-a197, October-2026, Available :http://www.jetir.org/papers/JETIR2610021.pdf

"Audio Visual Emotion Detection System", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 10, page no. ppa192-a197, October-2026, Available at : http://www.jetir.org/papers/JETIR2610021.pdf