Real-Time Detection of Multilingual Audio Distress Signals Using Compact CNN and MFCC Features in Noisy Environments

Key Points

• The aim is to develop a method for real-time detection of distress signals in multilingual audio environments.
• Used a compact convolutional neural network (CNN) for signal processing.
• Extracted Mel-frequency cepstral coefficients (MFCC) as audio features.
• Tested the model in various noisy environments to assess performance.
• Achieved effective detection rates of distress signals across multiple languages.
• Demonstrated robustness of the model in different noise levels.
• Showed improvements in detection speed compared to traditional methods.