AcoustiVision is a distributed environmental monitoring system that combines acoustic analysis and AI-powered vision to provide comprehensive real-time insights into indoor spaces. The system uses multiple BLE sensor nodes and a centralized Raspberry Pi dashboard for data collection, visualization, and analysis.

• Multi-Modal Sensing: Combines sound pressure level (SPL) monitoring with AI person detection
• Distributed Architecture: Multiple independent sensor nodes communicate via BLE
• Real-Time Visualization: Live graphs and statistics on an intuitive dashboard
• Automatic Data Logging: All sensor data timestamped and saved to CSV
• Smart Reconnection: Robust connection management with automatic recovery

┌─────────────────────────────────────────────────────────────┐
│                    Raspberry Pi Dashboard                    │
│              (Central Data Collection & Display)             │
│                                                              │
│  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐     │
│  │   Live SPL   │  │  Occupancy   │  │ Connection   │     │
│  │   Display    │  │   Counter    │  │     Log      │     │
│  └──────────────┘  └──────────────┘  └──────────────┘     │
│                                                              │
│  ┌────────────────────────────────────────────────────┐    │
│  │        Real-Time Graphs & Statistics               │    │
│  └────────────────────────────────────────────────────┘    │
└─────────────────────────────────────────────────────────────┘
                           │
                    BLE Connection
                           │
          ┌────────────────┴────────────────┐
          │                                 │
    ┌─────▼─────┐                    ┌─────▼─────┐
    │  SPL Node │                    │Vision Node│
    │(MG24 + Mic)                    │(ESP32-C3) │
    │           │                    │  + Grove  │
    │  Acoustic │                    │  AI V2    │
    │ Monitoring│                    │  Camera   │
    └───────────┘                    └───────────┘

• Hardware: Seeed Studio XIAO MG24 + Analog Microphone
• Function: A-weighted sound pressure level measurement
• Range: 30-120 dBA
• Update Rate: 2 Hz (500ms)
• BLE Service: Custom SPL measurement service
• Output: Float (dBA value)

• Hardware: Seeed Studio XIAO ESP32-C3 + Grove Vision AI V2
• Function: Real-time person detection and counting
• AI Model: SSCMA PeopleNet
• Update Rate: 1 Hz (1 second)
• BLE Service: Custom occupancy service
• Output: Integer (person count)

• Large Displays: Easy-to-read current values for both sensors
• Live Graphs: Scrolling time-series plots (last 100 samples)
• Statistics: Min/Max/Average calculations for SPL data
• Status Indicators: Clear connection health visualization

• Automatic CSV Export: All data timestamped and logged
• Synchronized Data: SPL and occupancy data aligned by timestamp
• Easy Analysis: CSV format compatible with Excel, Python, R

• Real-Time Log: See exactly what's happening with connections
• Auto-Reconnect: Automatic recovery from disconnections
• Health Monitoring: Visual indicators for connection quality
• Detailed Diagnostics: Error messages and troubleshooting info

• Raspberry Pi (Model 3B+, 4, or 5)
  • Built-in Bluetooth or USB BLE adapter
  • Running Raspberry Pi OS (Bullseye or newer)
  • Display (HDMI monitor or touchscreen)
  • Minimum 2GB RAM recommended

• Seeed Studio XIAO MG24 (Sense)
• Analog microphone (built-in or external)
• USB-C cable for power/programming

• Seeed Studio XIAO ESP32-C3
• Grove Vision AI V2 module
• Grove cable for I2C connection
• USB-C cable for power/programming

• USB power adapters (stable 5V supply recommended)
• Enclosures for sensor nodes
• Tripods or mounting hardware

# Update system
sudo apt update && sudo apt upgrade -y

# Install required packages
sudo apt install python3-pip python3-tk bluetooth bluez

# Enable Bluetooth
sudo systemctl enable bluetooth
sudo systemctl start bluetooth

pip3 install bleak matplotlib pandas

requirements.txt:

bleak>=0.21.0
matplotlib>=3.5.0
pandas>=1.3.0

• Arduino IDE with Seeed XIAO MG24 board support
• Libraries:
  • SilabsMicrophoneAnalog
  • ArduinoBLE
  • Custom SPL_Meter class

• Arduino IDE with ESP32 board support
• Libraries:
  • Seeed_Arduino_SSCMA
  • ESP32 BLE (native)
  • Wire (I2C)

# Open Arduino IDE
# Load: acousticNode.ino along with SPL_Meter.h and SPL_Meter.cpp
# Select Board: "Seeed XIAO MG24 (Sense)"
# Upload

# Open Arduino IDE
# Load: aiVisionNode.ino
# Select Board: "XIAO_ESP32C3"
# Upload

# Create project directory
mkdir ~/acoustivision
cd ~/acoustivision/dashboard

# Install dependencies
pip3 install -r requirements.txt

cd ~/acoustivision
python3 environmental_dashboard.py

Watch the Connection Log panel:

• ✓ Devices should appear in scan
• ✓ Connections should succeed
• ✓ Data should start flowing

┌─────────────────────────────────────────────────────────┐
│  AcoustiVision Dashboard          Status: 2/2 Connected │
├──────────────┬──────────────────────────────────────────┤
│              │                                          │
│  SPL METER   │                                          │
│              │         SOUND LEVEL GRAPH                │
│   65.2 dBA   │                                          │
│  Connected ✓ │                                          │
│              ├──────────────────────────────────────────┤
│  Min: 45.1   │                                          │
│  Max: 78.3   │         OCCUPANCY GRAPH                  │
│  Avg: 62.7   │                                          │
│              │                                          │
├──────────────┤                                          │
│              │                                          │
│ VISION NODE  │                                          │
│              │                                          │
│   3 people   │                                          │
│  Connected ✓ │                                          │
│              │                                          │
├──────────────┤                                          │
│ CONNECTION   │                                          │
│     LOG      │                                          │
│              │                                          │
│ [12:34:56]   │                                          │
│ SPL: 65.2 dBA│                                          │
│ [12:34:57]   │                                          │
│ Vision: 3    │                                          │
└──────────────┴──────────────────────────────────────────┘

AcoustiVision automatically creates CSV log files:

Format: sensor_data_YYYYMMDD_HHMMSS.csv

Example: sensor_data_20251108_143530.csv

Contents:

Timestamp,SPL_dBA,People_Count
2025-11-08 14:35:30.123,65.2,3
2025-11-08 14:35:30.623,64.8,3
2025-11-08 14:35:31.123,66.1,2

Check scan results in Connection Log

1. Verify sensors are powered and running
2. Check LED indicators on boards
3. Confirm device names match:
   • SPL Meter: SPL_Meter
   • Vision Node: AIVisionNode

# Manual BLE scan
sudo hcitool lescan
# Look for your device names

"Connection timeout" errors

1. Restart Bluetooth service:

   sudo systemctl restart bluetooth

2. Check Bluetooth permissions:

   sudo usermod -a -G bluetooth $USER
   # Logout and login

3. Reduce distance between devices (< 5 meters)

4. Power cycle sensor nodes

Devices keep dropping

1. Power Issues: Use wall adapters, not laptop USB
2. Interference: Move away from WiFi routers
3. Distance: Keep sensors within 5-10 meters
4. Reduce Update Rate:

   // In Arduino code
   #define BLE_UPDATE_INTERVAL 1000  // Slower = more stable

1. Test each sensor individually (power off one)
2. Check UUIDs match between firmware and dashboard
3. Verify characteristic formats:
   • SPL: Float (4 bytes)
   • Vision: Uint8 (1 byte)

Edit acoustivision_dashboard.py (lines 17-23):

# SPL Meter Configuration
SPL_DEVICE_NAME = "YOUR_SPL_NAME_HERE"

# Vision Node Configuration
VISION_DEVICE_NAME = "YOUR_VISION_NAME_HERE"

# Line 27-30
SCAN_TIMEOUT = 15.0          # Scan duration
RECONNECT_DELAY = 5.0        # Wait between reconnects
CONNECTION_TIMEOUT = 20.0    # Connection attempt timeout

# Line 33
MAX_DATA_POINTS = 100  # Points shown in graphs (50-200 recommended)

• Correlate noise levels with occupancy
• Optimize HVAC based on actual room usage
• Energy efficiency monitoring

• Environmental psychology studies
• Acoustic comfort analysis
• Occupancy pattern research
• Building performance evaluation

• Meeting room utilization
• Noise pollution monitoring
• Workspace optimization
• Productivity environment studies

• Unusual occupancy detection
• Emergency evacuation monitoring
• After-hours activity tracking
• Intrusion detection via noise

• Patient room monitoring
• Hospital noise management
• Occupancy tracking for infection control
• Staff activity analysis

import pandas as pd
import matplotlib.pyplot as plt

# Load data
df = pd.read_csv('sensor_data_20251108_143530.csv')
df['Timestamp'] = pd.to_datetime(df['Timestamp'])

# Calculate statistics
print("SPL Statistics:")
print(df['SPL_dBA'].describe())

print("\nOccupancy Statistics:")
print(df['People_Count'].value_counts())

# Plot correlation
plt.figure(figsize=(10, 5))
plt.scatter(df['People_Count'], df['SPL_dBA'], alpha=0.5)
plt.xlabel('People Count')
plt.ylabel('SPL (dBA)')
plt.title('Noise Level vs Occupancy')
plt.show()

# Hourly averages
df['Hour'] = df['Timestamp'].dt.hour
hourly_avg = df.groupby('Hour').agg({
    'SPL_dBA': 'mean',
    'People_Count': 'mean'
})

# Peak times
peak_occupancy = df.loc[df['People_Count'].idxmax()]
print(f"Peak occupancy: {peak_occupancy['People_Count']} at {peak_occupancy['Timestamp']}")

# Quiet periods (< 50 dBA and 0 people)
quiet_periods = df[(df['SPL_dBA'] < 50) & (df['People_Count'] == 0)]
print(f"Quiet time: {len(quiet_periods)} samples ({len(quiet_periods)/len(df)*100:.1f}%)")

Create systemd service:

sudo nano /etc/systemd/system/acoustivision.service

[Unit]
Description=AcoustiVision Environmental Monitor
After=bluetooth.target graphical.target

[Service]
Type=simple
User=pi
Environment=DISPLAY=:0
WorkingDirectory=/home/pi/acoustivision
ExecStart=/usr/bin/python3 /home/pi/acoustivision/acoustivision_dashboard.py
Restart=on-failure
RestartSec=10

[Install]
WantedBy=graphical.target

Enable:

sudo systemctl daemon-reload
sudo systemctl enable acoustivision.service
sudo systemctl start acoustivision.service

# Enable VNC
sudo raspi-config
# Interface Options → VNC → Enable

# Access from another computer
# VNC Viewer: raspberrypi.local:5900

Deploy multiple sensor sets:

# Room 1
SPL_DEVICE_NAME = "SPL_Room1"
VISION_DEVICE_NAME = "Vision_Room1"

# Room 2
SPL_DEVICE_NAME = "SPL_Room2"
VISION_DEVICE_NAME = "Vision_Room2"

Enable verbose logging:

# At start of main()
import logging
logging.basicConfig(level=logging.DEBUG)

We welcome contributions! Areas for improvement:

• Additional sensor types (temperature, humidity, CO2)
• Mobile app companion
• Cloud data sync
• Alert/notification system
• Machine learning analysis
• Web-based dashboard
• Multi-location aggregation

MIT License - See LICENSE file for details

• Seeed Studio for XIAO and Grove Vision AI v2 hardware platforms
• Bleak Python BLE library
• SSCMA AI framework
• Arduino, Silicon Labs and ESP32 communities