Your Smart Audio Companion for Environmental Awareness
soundAware.Presentation.mp4
soundAware is our 2-day Hackathon project! An Android app that transforms your phone into a smart audio detector, helping you stay aware of your surroundings through visual and vibration alerts. Whether you're deaf, hard of hearing, wearing headphones, or simply need to monitor your environment, soundAware acts as your personal sound assistant.
- Recognizes hundreds of different sounds including vehicles, alarms, voices, animals, and more
- Real-time audio analysis powered by advanced AI technology
- Shows what sounds are happening around you
- Pinpoints whether sounds are coming from your left, right, or center
- Visual arrows and color indicators guide you to the sound source
- Perfect for locating important sounds in busy environments
- 🟥 URGENT (Red): Immediate attention needed - loud, close sounds
- 🟧 APPROACHING (Orange): Sounds getting closer or louder
- 🟩 DISTANT (Green): Sounds detected but far away
- ⬜️ STABLE (Gray): Steady background sounds
- Different vibration patterns for different alert levels
- Feel urgent sounds through haptic feedback
- Customizable vibration intensity
- Large, easy-to-read display with high contrast colors
- Real-time sound names and confidence levels
- Clean, accessible design that works for everyone
- Deaf and Hard of Hearing Community: Stay connected to your audio environment
- People with Sensory processing issues: Be aware of important sounds without sensory overload
- Security Awareness: Know when vehicles, people, or alarms are nearby
- Anyone: Who wants better awareness of their surroundings
soundAware recognizes a wide variety of sounds including:
- 🚗 Vehicles: Cars, trucks, motorcycles, sirens
- 🔔 Alarms: Fire alarms, security alarms, phone rings
- 👥 People: Speech, footsteps, crying, laughter
- 🐕 Animals: Dogs barking, cats meowing, birds chirping
- 🏠 Home: Doors closing, appliances, music
- 🚨 Emergency: Sirens, horns, urgent alerts
Audio Processing Pipeline:
- Listen: Continuous audio capture at 16kHz sample rate using Android's AudioRecord API
- Pre-processing: YAMNet model handles internal audio windowing (0.96s frames with 0.48s overlap)
- Feature Extraction: YAMNet model performs mel spectrogram conversion internally
- AI Classification: YAMNet model performs real-time inference for 521 sound classes
- Spatial Analysis: Stereo audio processing to determine left/right channel dominance for sound localization
- Proximity Detection: Dynamic volume threshold analysis and RMS-based distance estimation
- Alert Generation: UI updates with haptic feedback patterns
Key Technologies:
- Machine Learning: YAMNet-based model optimized for mobile deployment (TensorFlow Lite)
- Audio Processing: Hardware-accelerated inference with NNAPI support
- Privacy-First: All processing happens locally on-device - no audio data leaves your phone
User Experience Flow:
- Listen: SoundAware continuously monitors audio through your phone's microphone
- Analyze: Advanced AI instantly identifies and classifies sounds
- Locate: Determines if sounds are coming from left, right, or center
- Alert: Shows visual indicators and provides vibration feedback
- Track: Monitors if sounds are getting closer, moving away, or staying steady
- Best First-Time Hacker Team
soundAware - Making sound Visual and Haptic