Skip to content

SunbirdAI/TalkToSema-Device-Documentation

Repository files navigation

Firmware and Software Overview

Repository: voice-feedback-device-documentation

This section documents the firmware implementation for the first-generation voice feedback device, as maintained in the project GitHub repository.


Core Libraries Used

  • TinyGSM – LTE modem communication
  • SPI – Peripheral communication interface
  • SD – Local SD card storage
  • TMRPCM – Audio recording and playback in WAV format
  • EEPROM – Persistent storage of counters and device state
  • ARDUINO-TIMER – Task scheduling and low-power timing

Device Operation Logic

  1. Device remains in low-power sleep mode by default
  2. User presses button → device wakes up
  3. Records a voice feedback audio file (.wav)
  4. Stores file on SD card
  5. Increments file counter
  6. When 5 files are recorded:
    • LTE modem activates
    • All 5 files are uploaded to the citizen feedback portal
    • Counter resets to 1
  7. Device returns to sleep mode

Device Architecture

*Figure: Sema Audio Feedback Device Architecture*

Sema Audio Feedback Device Versioning and Comparison

This document summarizes the evolution of the Sema Audio Feedback Devices across four major iterations. It highlights the hardware choices, functional improvements, power performance, and deployment considerations to support decision‑making on which device version is best suited for specific field deployments.


Objective

To provide a clear comparison of Sema device versions developed over time, showing technological progress, feature improvements, and deployment trade‑offs for supervisors and project stakeholders.


Device Version Summaries

Device Version 1.0 – Initial Prototype

Purpose: First proof‑of‑concept for collecting citizen audio feedback and transmitting to the portal.

Key Features

  • Main controller: Arduino Mega Pro Mini
  • Audio format: WAV recording
  • Transmission: LTE modem upload to citizen feedback portal
  • Maximum audio length: 15 seconds
  • Power: 2 lithium batteries (series or parallel configuration)
  • Runtime: ~2 days
  • Audio quality: Raw microphone input, no background noise suppression

Pros

  • Functional end‑to‑end audio capture and upload
  • Simple and low-cost design

Cons

  • Short battery life
  • Limited recording duration
  • No noise suppression

Device Architecture

*Figure: Sema Audio Feedback Device Version 1.0*

Device Version 1.2 – Enhanced Audio Quality and Runtime

Purpose: Improved upon v1.0 with longer recording and better audio clarity.

Key Features

  • Main controller: Arduino Mega Pro Mini
  • Audio format: WAV recording
  • Transmission: LTE modem upload to citizen feedback portal
  • Maximum audio length: Up to 1 minute (can stop early when user finishes)
  • Microphone: Noise‑cancellation microphone
  • Power: 3 lithium batteries in series
  • Runtime: ~7 days
  • Audio quality: Improved with background noise suppression

Pros

  • Better audio clarity
  • Longer battery life
  • Longer flexible recording duration

Cons

  • Increased power complexity
  • Still dependent on LTE connectivity
Device Architecture Device Architecture
Version 1.2.1 Version 1.2.2

Device Version 2 – Offline Local Storage Device

Purpose: Designed for deployments without continuous internet connectivity.

Key Features

  • Main controller: Arduino Nano
  • Audio format: WAV recording
  • Transmission: No cloud upload; files stored locally
  • Maximum audio length: Up to 1 minute (early stop supported)
  • Microphone: Noise‑cancellation microphone
  • Power: 3 lithium batteries in series
  • Runtime: ~14 days

Pros

  • Longest battery life among Arduino-based versions
  • Does not require network connectivity
  • High audio quality

Cons

  • Requires manual retrieval of stored audio files
  • No real‑time monitoring

Device Architecture

*Figure: Sema Audio Feedback Device Version 2.0*

Device Version 3 – Smart IoT Enabled Device

Purpose: Transition to embedded Linux for advanced processing and connectivity.

Key Features

  • Main controller: Raspberry Pi Zero 2W
  • Audio input: USB microphone
  • Transmission: Raspberry Pi 4G HAT with USB attachment
  • Audio handling: On‑device recording and processing
  • Runtime: ~4 hours

Pros

  • Supports advanced processing and future ML integration
  • Flexible software environment
  • Direct cloud connectivity

Cons

  • Short battery life
  • Higher power consumption
  • Increased system complexity

Device Architecture

*Figure: Sema Audio Feedback Device Version 3.0*

Comparative Summary Table

Feature Version 1.0 Version 1.2 Version 2 Version 3
Main Controller Arduino Mega Pro Mini Arduino Mega Pro Mini Arduino Nano Raspberry Pi Zero 2W
Audio Format WAV WAV WAV WAV
Max Recording Length 15 sec 1 min 1 min Configurable
Audio Upload LTE to portal LTE to portal Local storage only 4G Cloud upload
Microphone Standard Noise‑cancellation Noise‑cancellation USB Microphone
Noise Suppression No Yes Yes Depends on software
Power Source 2 Li batteries 3 Li batteries 3 Li batteries Li battery pack
Runtime ~2 days ~7 days ~14 days ~4 hours
Connectivity Requirement LTE LTE None 4G
Deployment Mode Online Online Offline Online / Smart IoT

Deployment Recommendations

Short-term citizen feedback pilots (connected areas):

  • Version 1.2 – Reliable connectivity and good audio quality

Remote or off-grid deployments:

  • Version 2 – Long battery life and offline storage

Advanced sensing and real-time monitoring:

  • Version 3 – Supports future ML, dashboards, and OTA updates

Device Architecture

*Figure: Sema Audio Feedback Device Deployed*

Evolution Highlights

  • Progressive improvement in battery efficiency
  • Introduction of noise‑cancellation microphones
  • Transition from microcontroller to embedded Linux
  • Shift from short recordings to flexible user-controlled recordings
  • Gradual integration toward smart connected sensing platforms

Future Directions

  • Power optimization for Raspberry Pi platform
  • Hybrid online/offline synchronization
  • On-device speech-to-text or classification
  • Solar-assisted charging

Circuit Schematics

You can view the circuit schematics for each device version below:


Power Supply and Charging Architecture

The Sema Audio Feedback Devices support flexible power input for both grid-connected and off-grid deployments.

Battery Configurations

  • Series configuration: 12.6V DC charging (12V adapter or solar)
  • Parallel configuration: 5V DC charging (low-power systems)

Integrated Buck-Boost Regulation

The PCB includes an on-board buck-boost converter that:

  • Accepts solar panels or DC adapters (up to 20V max input)
  • Automatically regulates voltage for battery charging and device operation
  • Enables direct solar panel connection without external regulation

Power Input Options

  • Mains: 12V DC adapter
  • Solar: Panel (≤20V) via onboard regulator
  • Battery: Internal lithium pack

SEMA Device Cost Budget

SN Item Name Description Qty Unit Cost (UGX) Total Cost (UGX)
1 Lithium batteries 2600mAh 3 5,000 15,000
2 Microcontroller Mega Pro Mini 1 80,000 80,000
3 4G Modem (inbuilt GPS) SIM7600 1 140,000 140,000
4 Lithium charger 3S charger 1 20,000 20,000
5 Recording module MAX9814 1 20,000 20,000
6 SD card reader Read SD card 1 10,000 10,000
7 SD card 1GB 1 15,000 15,000
8 Switch Power device on or off 1 3,000 3,000
9 RTC DS3231 1 12,000 12,000
10 Voltage step-down 1 Circuit step-down 1 20,000 20,000
11 Switching transistor TIP122 1 1,500 1,500
12 Switching resistor 1K Ohm 2 500 1,000
13 Power block diodes 2N1001 2 1,000 2,000
14 Header pins Component slot 1 5,000 5,000
15 Glue stick Assembly 1 5,000 5,000
16 RTC battery RTC power 1 3,000 3,000
17 PCB PCB board 1 30,000 30,000
Total (UGX) 382,500
Total (USD) ~103.21

Supplier: Bbiri Centre


Author: Joel T. Muhanguzi
Project: Sema Audio Feedback Hardware Evolution

About

TalktoSema Feedback Device Documentation

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages