Concept
Inspiration
Limitations/ How the path changed
Testing environments for Audio modification - Python/ Touchdesigner/ Arduino

Swarna Manjari

•

Heba Elidrisi

•

Beste Cebeci

Published March 6, 2026

Seeding Memories

A secret memory encrypting system merging real-time human audio with the living pulses of a tree to build a private archive of memories.

IntermediateWork in progress31

Things used in this project

Hardware components

Breadboard (generic)

Barduino

Jumper wires (generic)

9V battery (generic)

Analog Devices EMG Muscle Sensor AD8226

5 mm LED: Green

Fiber Optic Cable, Duplex

Circular Connector Cable Seal, Heat Shrinkable Sealing Boot

Headset with Microphone

Software apps and online services

Touch Designer

Arduino IDE

Python

Hand tools and fabrication machines

Soldering iron (generic)

Wire Stripper & Cutter, 32-20 AWG / 0.05-0.5mm² Solid & Stranded Wires

Story

Concept

What if you could whisper your secrets to a tree that promises to store it forever in the depths of it's heart in a language no one else understands? What if you could go back to hear it whisper it back to you again?

This is where our idea for "Seeding Memories" was born. We wanted to merge the sensory data from the electrical impulses of a living tree and use it to encrypt our secret message; an act of trusting the tree with our personal memories. The tree, using the data, translates it into a language that distorts our message keeping it secret.

Inspiration

We were inspired by the guardian 'Tree of Souls' from Avatar as a space for storing community memory. In our version we wanted it to store our personal data in a way that is encrypted in a secret shared language between tree and human, creating an intimacy between two different intelligences.

Sketches (Development of Idea)

While brainstorming, we initially we wanted to develop the idea of a seed as a microphone to speak your memory into; for it to receive the audio message and plant it into its memory. Then, it would play the encrypted message out to you. The seed would connect to the rest of the artifact that holds the EMG sensor electrode pads and the barduino board, bringing it all together in one object.

Working pathway

1. Making the EMG sensor work. Reading the data to be accurate.

2. Check playing an audio file on an amplifier.

3. See if you can merge the audio data with the sensor data.

4. Play it back using a speaker.

1 / 3

Limitations/ How the path changed

1. Challenging to read the sensor data

The EMG sensor is a very sensitive sensor that detects microvoltaic changes. Thus, the amount of voltage it receives is crucial. This was a dual rail sensor requiring voltage input from two batteries. Thus connecting the setup was more challenging than we thought it would be. To read the accurate reading from the muscle, we had to check the placement of the electrodes, and the connections of the sensor again and again. Any loose contact affected the sensor read in a large magnitude.

reading from sensor

Things to keep in mind:

Connecting the laptop to power and reading the sensor affects its values.

2. Reading a plant using the sensor

The microvolts from a human muscle contraption varies in magnitude compared to the plant signals. Thus, the readings were not significantly influential.

3. Trying the amplifier

1 / 2 • trying three versions

We tried to use a DFplayer robot mini to play some audio and use the space in the SD card to store the received message so that it could be replayed in the future. However after multiple attempts, we couldn't connect the player to the Barduino. Thus we decide to change the environment.

Testing environments for Audio modification - Python/ Touchdesigner/ Arduino

1. For trying to modify the live audio data, we initially used a python code to distort it.

2. We had to make the sensor data communicate with the python environment.

3. However the coding didn't feel dynamic enough to modulate/play with the audio.

4. So, we decided to switch to TouchDesigner.

workflow

Since we were not able to program the DFplayer, we decide to use a headphone and a microphone, instead of an external speaker and a mic.

audio setup

Designing the Seed

We wanted the seed to mimic the glow, so we used fibre optic cables connected to an led light. We bound the cables with dark thermal shrinkable sealing boot to cover the led and make it reflect only on the tips.

1 / 3

Testing Lights

We connected the pulsing led to reflect the sensor data being in an active state of reading.

Lights, seeds, action

Bringing them all together

1 / 2

Code

const int emgPin = 7;
const int ledPin = 9;


unsigned long previousMillis = 0;
unsigned long interval = 50;

unsigned long startTime;

int brightness = 0;
int fadeAmount = 2;

void setup() {
  pinMode(ledPin, OUTPUT);
  Serial.begin(115200);
  
  startTime = millis(); // record when program starts
}

void loop() {

  int emgValue = analogRead(emgPin);
  Serial.println(emgValue);

  // wait 20 seconds before using EMG to control LED
  if (millis() - startTime < 20000) {
    analogWrite(ledPin, 0);   // LED off during calibration
    return;
  }

  // after 20 seconds, EMG controls the LED timing
  interval = map(emgValue, 0, 200, 80, 15);
  interval = constrain(interval, 20, 120);

  unsigned long currentMillis = millis();

  if (currentMillis - previousMillis >= interval) {
    previousMillis = currentMillis;

    brightness += fadeAmount;

    if (brightness <= 0 || brightness >= 200) {
      fadeAmount = -fadeAmount;
    }

    analogWrite(ledPin, brightness);
  }
  delay(50);   
}

Seeding Memories