Skip to the content
2 layers45 × 22.3 mmFlexsonic 2, the printed circuit board in 3D

plaintext Flexsonic ├── .obsidian/ # Notes (Obsidian workspace) │ └── Pre-requisites.md # Basic notes │ ├── audio/ # All audio files in use │ ├── build/ # ESP-IDF build output │ ├── datacollection/ # Raw data logging +…

Size
45 × 22.3 mm
Copper layers
2
Thickness
1.6 mm
Parts
12
Nets
18
Pads
32
Tracks, vias
37, 6
Pours
1
Routed
Yes
KiCad
KiCad 9.0 (file 20241229)
Created
26 Jun 2025
Updated
14 Oct 2025

Overview

Flexsonic 2 is an open-source printed circuit board by Shri-2112, from the Shri-2112/Flexsonic repository on GitHub. It is a 2-layer board of 45 × 22.3 mm with 12 parts on 18 nets, routed with 37 tracks and 6 vias.

Its parts include 6 connectors, 1 resistor, 1 diode and 1 switch.

README

README.md

FlexSonic – Smart Glove for Gesture-to-Speech Communication

A wearable assistive device that translates hand gestures into audible speech using flex sensors, IMU, ESP32, and DFPlayer Mini.


Table of Contents


About the Project

Aim

FlexSonic aims to bridge the communication gap for speech-impaired individuals by converting hand gestures into audible phrases.

Description

The glove uses flex sensors (to detect finger bending) and an MPU6050 IMU (to capture orientation). Sensor data is processed on an ESP32, which runs a simple ML model (K-Means clustering) to classify gestures. The recognized gesture is mapped to a pre-recorded audio file on a DFPlayer Mini, producing real-time speech output through a speaker.

Tech Stack

Hardware:

  • ESP32
  • Flex Sensors
  • MPU6050
  • DFPlayer Mini
  • Speaker
  • SD Card

Software:

  • Arduino IDE / PlatformIO (ESP32 firmware)
  • C / Embedded C for driver code
  • Python (for ML model training & testing)
  • ML Libraries: scikit-learn

Frameworks/Tools:

  • FreeRTOS (ESP32 task scheduling)
  • I²C, UART communication protocols

File Structure

plaintext
Flexsonic
├── .obsidian/                  # Notes (Obsidian workspace)
│   └── Pre-requisites.md       # Basic notes
│
├── audio/                      # All audio files in use
│
├── build/                      # ESP-IDF build output
│
├── data_collection/            # Raw data logging + visualization
│   ├── data.py                 # Data logging script
│   ├── data.txt                # Raw dataset (text format)
│   ├── graph.py                # Graphical representation of dataset
 
├── data_processed/             # Processed dataset
│   ├── data_with_clusters.csv  # Clustered dataset
│   ├── gesture_labeled.csv     # Labeled dataset
│   └── gesture_parsed.csv      # Parsed dataset
│
├── main/                       # ESP32 firmware (C code)
│   ├── 1_flex.c
│   ├── 2_mpu.c
│   ├── 3_mpu_and_flex.c
│   ├── 4_sentence_gesture.c
│   └── 5_numbers_gesture.c

Getting Started

Prerequisites

text
- **ESP-IDF v4.0+** (or Arduino IDE with ESP32 boards installed)  
- **Python 3.8+**  

Required libraries (Arduino/ESP32):

  • esp32 board support package
  • Adafruit_MPU6050
  • DFRobotDFPlayerMini

⚙️ Installation

Clone the repo:

bash
git clone https://github.com/your_username/flexsonic.git
cd flexsonic

Usage

Follow these steps to run and test the project:

  1. Connect Hardware

    • Attach all flex sensors to the ESP32.
    • Connect the MPU6050 via I2C.
    • Insert the SD card into the DFPlayer Mini.
    • Connect a small speaker to the DFPlayer Mini output.
  2. Upload Code

    • Open the project folder in VS Code with PlatformIO or Arduino IDE.
    • Select the ESP32 board and correct COM port.
    • Upload the firmware.
  3. Run the Project

    • Power on the ESP32 (via USB or battery).
    • Perform different hand gestures.
    • The ESP32 will recognize gestures using sensor data.
    • DFPlayer Mini will play the corresponding pre-recorded audio.
  4. Add New Gestures

    • Update the gesture mapping logic in main.cpp.
    • Add new audio files (.mp3) to the SD card in sequential numbering (e.g., 0001.mp3, 0002.mp3).
    • Re-flash the code to include new mappings.
  5. Testing

    • Open the Serial Monitor at 115200 baud rate.
    • Debug sensor readings and check recognized gestures.
text

Results & Demo

Results

  • Successfully translates predefined hand gestures into corresponding audio outputs.
  • Flex sensors + MPU6050 data are processed in real-time by the ESP32.
  • DFPlayer Mini plays pre-recorded audio files with minimal delay.
  • System can be extended by adding more gestures and audio mappings.
  • Achieved smooth serial debugging and stable performance in lab tests.
text

Demo

You can check out the project demo here:

text

Future Work

  • Gesture Expansion: Add more complex gesture combinations for a richer vocabulary.
  • Dynamic Phrase Generation: Integrate an ML model for real-time text-to-speech instead of fixed audio files.
  • Miniaturization: Design a compact, custom PCB to reduce size and improve portability.
  • Wireless Connectivity: Enable Bluetooth/Wi-Fi support for integration with smartphones or IoT devices.
  • Energy Optimization: Implement low-power modes for longer battery life.
  • Multilingual Support: Expand gesture-to-speech output across multiple languages.
  • AI Integration: Use advanced ML techniques (e.g., LSTM/RNN) for continuous gesture recognition.
  • Cloud/Edge Support: Store and update gesture mappings via a connected app or cloud server.

Troubleshooting

Here are some common issues and their solutions:

  • No audio output from DFPlayer Mini

    • Check if the SD card is properly formatted (FAT32).
    • Ensure audio files are named correctly (e.g., 0001.mp3, 0002.mp3).
    • Verify speaker connections and volume settings.
  • ESP32 not detecting flex sensors

    • Recheck wiring of ADC pins.
    • Calibrate sensor values in the code.
    • Test with a multimeter to confirm sensor resistance changes.
  • IMU (MPU6050) giving unstable readings

    • Ensure proper power supply (3.3V, not 5V).
    • Add a small delay/filtering in code to stabilize readings.
    • Check I2C connections (SDA, SCL).
  • Code not compiling

    • Verify required libraries are installed (DFRobotDFPlayerMini, Wire, etc.).
    • Ensure correct board and COM port are selected in Arduino IDE.
  • Gestures not recognized correctly

    • Increase training data for the ML model.
    • Adjust threshold values in the code.
    • Check if flex sensors are securely attached to fingers.

Contributors

This project was made possible with the efforts of:

Mentees

  • Janhavi Mokal
  • Shrinivas Pathak

Mentors

  • Bhavesh Phundhkar
  • Yash Suthar
  • Swanand Patil

Contributions are always welcome!
Feel free to fork this repo, create a branch, and submit a pull request.


Acknowledgment

A big thank you to:

  • SRA VJTI Mumbai for providing the academic environment to pursue this project.
  • Open-source communities whose libraries and resources made implementation easier.
  • Everyone who inspired, tested, and supported FlexSonic in its early stages.

Resources

Here are some references and tools that helped during development:

Parts

12
QtyValueFootprintReferences
2Conn_01x08Connector_PinSocket_2.54mm:PinSocket_1x08_P2.54mm_VerticalJ4, J7
1220Resistor_SMD:R_0805_2012MetricR1
15Vinput_DfConnector_JST:JST_PH_B2B-PH-K_1x02_P2.00mm_VerticalJ5
15voutputConnector_JST:JST_PH_B2B-PH-K_1x02_P2.00mm_VerticalJ2
1LEDLED_SMD:LED_0805_2012MetricD1
1LOGOlogos:sra_logo_small_negG***
1LOGOLogos:Lions_logoG***
1LOGOlogos:sra_skull_bigG***
1SW_SPSTButton_Switch_SMD:SW_SPST_PTS645Sx43SMTR92SW1
1SpeakerConnector_JST:JST_PH_B2B-PH-K_1x02_P2.00mm_VerticalJ6
1Uart_tx_rxConnector_JST:JST_PH_B2B-PH-K_1x02_P2.00mm_VerticalJ1

More from Shri-2112/Flexsonic

1
  • 2 layers45.9 × 44.6 mmFinal., the printed circuit board in 3D

    Final.

    Shri-2112/Flexsonic

    plaintext Flexsonic ├── .obsidian/ # Notes (Obsidian workspace) │ └── Pre-requisites.md # Basic notes │ ├── audio/ # All audio files in use │ ├── build/ # ESP-IDF build output │ ├── datacollection/ # Raw data logging +…

    40 parts

Similar boards

  • 2 layersNrf bga, the printed circuit board in 3D

    Nrf bga

    github-linguist/linguist

    Language Savant. If your repository's language is being reported incorrectly, send us a pull request!

    12 parts14kMIT
  • 2 layers66.9 × 74.5 mmDeskHop, the printed circuit board in 3D

    DeskHop

    hrvach/deskhop

    Fast Desktop Switching Device

    12 parts7.9kGPL-3.0
  • 2 layers40 × 15 mmSensors, the printed circuit board in 3D

    Sensors

    google/mozc-devices

    Circuit diagrams and firmware source code for Gboard DIY keyboards

    12 parts3kApache-2.0
  • 2 layers42.7 × 56.5 mmTle6240 driver, the printed circuit board in 3D

    Tle6240 driver

    rusefi/rusefi

    rusefi - GPL internal combustion engine control unit

    12 parts1.2k
  • 2 layers49 × 25 mm5v boost regulator, the printed circuit board in 3D

    5v boost regulator

    hackclub/OnBoard

    💡 Join 1,000 teenagers and make your first PCB with a $100 grant!

    12 parts915
  • 2 layers28.3 × 10.3 mmTap Photosensor PCB, the printed circuit board in 3D

    Tap Photosensor PCB

    VoronDesign/Voron-Tap

    Tap is a nozzle-based z-probe for the V2 and Trident printer designs. The entire toolhead moves to trigger an optical switch. Tap offers many advantages:

    12 parts716GPL-3.0
  • 2 layers25.4 × 70.4 mmTpmsniffer, the printed circuit board in 3D

    Tpmsniffer

    stacksmashing/pico-tpmsniffer

    A simple, very experimental TPM sniffer for LPC bus

    12 parts696

Questions

How big is the Flexsonic 2 PCB?

It measures 45 × 22.3 mm and has 2 copper layers, 1.6 mm thick.

What parts does the Flexsonic 2 use?

12 parts. The full list of its parts is on this page, with links to each part's symbol and footprint where PCBKit's library has them.

Where can I download the Flexsonic 2 KiCad files?

From the Shri-2112/Flexsonic repository on GitHub, at the commit this page was made from: the board is PCBs/Flexsonic_2.kicad_pcb and its schematic Schematics/Final..kicad_sch. PCBKit opens the board in the browser, with nothing to install.

Can I use the Flexsonic 2 design in my own project?

Its repository states no licence, so its author keeps all rights: ask Shri-2112 before reusing the design.