stardance has been extended another month! the new deadline is october 31 :)

You are browsing as a guest. Sign up (or log in) to start making projects!

weave

@weave

Joined August 18th, 2026

  • 3Devlogs
  • 2Projects
  • 0Ships
  • 0Votes
Open comments for this post

2h 33m 36s logged

Addition to day 2 or early day3:
Project overview: So I created this little AI which uses one of the model of groq and I called it E.V. Its an physical desktop assistant that listens to voice inputs from your microphone for my case its my airpods, it processes the question through cloud LLM, outputs responses through a 16x2 LCD screen, and speaks the answer back to you through text to speech.

How it works:
Python captures the audio from my mic using speech_recognition through google speech recognition API.

Then the text gets sent to groq (openai/gpt-oss-20b) and generate the actual answer. I set a strict limiter of tempature 0.0 so the AI can give me a super short and factual answer so it stays in the 30 character of the LCD.

Python then cleans up the string cutting any reasoning tags and other unnecessary things so the final answer comes out at 30 characters.

Python sends the cleaned text over USB serial (com5) to the arduino UNO to display the text on the 16x2 LCD screen, and make pyttsx3 to speak the response outloud.

Alot of problems came through when I did this project, the three most annoying ones were definitely getting the AI to give an answer in the 32 character limit of the LCD, the second one was definitely removing those thinking processes,why?
its because when it starts thinking it doesn’t seem to want to give me the final answer and is just stuck on thinking, Last problem is the text not showing up on the screen took me a whole half an hour to realise i put the jumper wire into the wrong digital pin.

The image below I tried to see if it could explain to me all of quantum physics in 30 words which is pretty funny, and is to show what the hardware limit is.

overall this project was extremely fun, if I had a plan to do this again but better next time I would definitely upgrade to rasberry Pi so all of the things like mic, speech engine and speaker can run on the board itself without needing it to plug into my laptop the whole time.

ill add another image after this post to show the build not just the terminal.

0
0
3
Open comments for this post

30m 11s logged

Day 2: so for today I built an ultrasonic sensor which triggers the servo motor. Alot of thing went wrong suprisingly, one of the things that went wrong was that I tried to put 25,000 in the pulseIn() instead of 25000 which made the timeout read 25, 000 instead of the 25000 I wanted. Not alot of thing went wrong with the building itself. I also found out how to make a 5V power rail on the breadboard so that all my components can get the power they need, just like how I did it with the GND! The video below shows how the built perform! If the video doesn’t play just know that if I put anything infront of the ultrasonic sensor it makes the little blue motor move its white arm 90 degrees.

0
0
6
Open comments for this post

33m 32s logged

Spent this session working on understanding how C++ and Arduino components work. Learnt how jumper wires can connect the common (-) together also learnt about the annoying issue of short circuits. For C++ I learned how Variable and Function works. I put together a pretty sweet mini project of a sound sensor to trigger these leds! not much for the first log sadly. (Day 1 this was built yesterday!)

0
0
19

Followers

Loading…