Real Time Spatial Sound Scene Analysis-AlertNet

This project introduces a groundbreaking approach to enhance the accuracy of acoustic scene classification and ensures user safety within music player applications. It provides a combined model of deep learning by merging Convolutional Neural Networks (CNN) with Bidirectional Long Short-Term Memory...

Full description

Saved in:
Bibliographic Details
Published in2024 International Conference on Advances in Computing, Communication and Applied Informatics (ACCAI) pp. 1 - 6
Main Authors Nagaraj, S, Malathy, R, Merlin Sheeba, G
Format Conference Proceeding
LanguageEnglish
Published IEEE 09.05.2024
Subjects
Accuracy
acoustic signal processing
Alerting system
Audio feature extraction
Bi-LSTM
CNN
Deep learning
Feature extraction
GFCC
MFCC
Performance evaluation
Predictive models
Scene classification
User experience
Online AccessGet full text

Cover

More Information
Summary:This project introduces a groundbreaking approach to enhance the accuracy of acoustic scene classification and ensures user safety within music player applications. It provides a combined model of deep learning by merging Convolutional Neural Networks (CNN) with Bidirectional Long Short-Term Memory (Bi-LSTM).The model uses Mel-Frequency Cepstral Coefficients (MFCC) and Gammatone Frequency Cepstral Coefficients (GFCC) features as feature extractors, and processes these features to predict scenes.Furthermore, this paper integrates the advanced scene classification model into music player applications. Users can enjoy their favourite music while the system continuously analyses the audio environment. When a potentially dangerous situation is detected, such as sirens, alarms, or other alerting sounds, the music playback is momentarily attenuated or paused, and users are promptly alerted to pay attention to their surroundings. Performance evaluation demonstrates 97 \% accuracy in a noisy environment and 98 \% accuracy in clean environment.The CNN-Bi-LSTM hybrid model, named AlertNet, excels in capturing both fine-grained and contextual audio information. Its seamless integration into music players enhances safety and user experience. Finally, it exemplifies the fusion of cutting-edge deep learning techniques with everyday technology to ensure safety and situational awareness for users, especially in scenarios where distractions are common.
ISBN:9798350389432
DOI:10.1109/ACCAI61061.2024.10601929