AudiAlert: Your ears, enhanced!

https://docs.google.com/document/d/1Dsop4wFZ51e7jIKgZiCBEeE9tVpui_RmxcrMsuZfON0/edit?usp=sharing

About AudiAlert AudiAlert is an innovative sound detection system that uses artificial intelligence to identify and analyze various sounds in real-time, enhancing safety and awareness in diverse environments.

Inspiration Our team was inspired by the potential of AI to help people with hearing disabilities, improve public safety and environmental monitoring. We realized that sound, often overlooked, carries crucial information about our surroundings, When I said sound, often overlooked, this really happened to us when we were on our trip to the Reed Bingham State Park, you go out to places like parks and go hiking to get yourself out from all the busy schedule you had, but nature had some other plans for us, on our chilling day, we suddenly started to hear weird rustling noises which we weren't even sure if it was really rustling or something dangerous like a bear, we were really scared to go look. Then being engineers we thought to ourselves how great it would be if we had something a tap away which tells us what exactly that is so that we can worry less and make our way out of danger. Hence is our inspiration that led us to create AudiAlert, a system that can "hear" and interpret sounds that humans might miss or misunderstand.

What it does Sound Detection: AudiAlert listens for specific sounds in the environment, such as hissing snakes, gunshots, alarms, and more. Real-time Analysis: Using convolutional Neural Networks (CNNs), the system converts audio inputs into spectrograms-visual representations of sound frequencies over time allowing for precise analysis. Instant Alerts: When a critical sound is detected, AudiAlert sends real-time notifications to users, enabling them to respond quickly to potential dangers. Adaptable Learning: The system continuously learns from new audio samples, improving its accuracy and expanding its ability to recognize additional sounds over time. User-Friendly Interface: AudiAlert features an intuitive interface that allows users to monitor sound alerts easily and access historical data for further analysis. Hence by combining cutting-edge AI technology with practical applications, AudiAlert empowers users to stay informed and safe in their environments.

How we built it We used Google's TM to train our model on various sound samples. Our system converts audio inputs into spectrograms for visual analysis. We implemented a CNN to classify these spectrograms accurately. We integrated a user-friendly interface for real-time alerts and sound classification.

Challenges we ran into Data Collection: Gathering a diverse set of high-quality audio samples was challenging. Model Accuracy: Fine-tuning the model to reduce false positives while maintaining sensitivity. Real-time Processing: Optimizing our system to analyze sounds quickly without lag.

Accomplishments that we're proud of We take pride in several key achievements during the development of AudiAlert Accurate Sound Classification: Successfully developed a model that identifies critical sounds like hissing snakes and gunshots. Real-time alerts: Implemented an instant notification system that enhances user safety and awareness. User-Friendly Interface: Created an intuitive interface for easy monitoring of sound alerts and historical data. Diverse Dataset: Compiled a comprehensive audio dataset that improved model accuracy and serves as a resource for future enhancements. Skill Development: Advanced our technical expertise in machine learning, audio processing, and software development.

What we learned Audio signal Processing: We learned how to analyze and interpret audio signals, understanding the significance of frequency and amplitude in sound recognition. Machine Learning fundamentals: Working with CNNs deepened our knowledge of machine learning algorithms, particularly in how they can be applied to sound classification. Teachable Machine: We explored Google's Teachable Machine, discovering its capabilities for training AI models with minimal coding, which streamlined our development process. Collaboration and Problem-Solving: Working as a team fostered collaboration and creative problem-solving skills, as we navigated challenges and iterated on our ideas together.

What's next for AudiAlert: Your ears, enhanced!: We aim to expand AudiAlert's capabilities by: Increasing our sound database. Implementing edge computing for faster processing. Developing mobile applications for wider accessibility.

Built With

• cnn
• fft
• spectrograms
• teachable-machines
• tensorflow