Help with an idea (doorbell problem)

Hey, I have been programming for 6 years now and am pretty advanced, but I only used an Arduino one time, and am not very confident in my hardware and wiring abilities.

I live in a student shared apartment, and we have a doorbell that rings when someone presses either the button in the entrance (2 floors down) or the one on our door. The problem is, the doorbell is very quiet, and we don’t hear it well. I thought it was a fun idea to make some sort of program to solve this problem.

I was thinking of some way in which the Arduino can recognize the sound of the doorbell, and when it does (ergo the doorbell rings), it should send a notification to a self-made app on our phone. (Let me take care of the app :stuck_out_tongue: )

How can I build such a contraption (what hardware do I need) and is there an easier method to solving this problem?

Thank you for your help/suggestions!

Have you considered replacing the doorbell by a more noisy one? :slightly_smiling_face:

Best Regards,
Johi.

That would be the boring solution :nerd_face:

An actual great way to get into ML with a 32bit MCU.

There is a Arduino, I think its a Nano 32 something or another with a built in sound sensor. I use ESP32’s and would have to add in a sound module for such a project.

With ML, machine learning, you would create a project that would record in digital, the sound of the doorbell.

Once you got the data model of your doorbell sound it will be a choice of either TensorFlow Lite or Linear Regression, The Nano 32 or a ESP32 can do 32 and 64 bit floating point linear regression. Now, when the sounds come in on the mic, the sound stream can be compared against the trained model and “that a doorbell”.

Such a project will teach you a lot.

1 Like

Amazing, thanks.

I am actually really interested in ML and would love to get into it. I will look into it!

I’m working on using a ESP32-CAM, got that running, taking pics, gray scale, to develop a training model to ID a bird., using Linear Regression. Right now I can ID a bird with .4 to .63 correlation. Just got to work on my training model some more.

Th ESP32-CAM sends its pic to a RPi, which is running the Linear Regression on the image. Whiles a ESP32 will work, the ESP32 CAM is taxed as it is to take the pic, and ftp it to the RPi for processing.

I am sure a ESP32 with a WaveShare variable gain sound module or a NANO32 would do the job.

Nice, that’s really cool.
Could you give me the name of the hardware that I should use? I kind of got lost with all the names.

ESP32 model S developer board.

Opops that is an Adafruit sound module.

sound module

A LinearRegression library, brush up on your normalizing of data.

This library is 64 bit and very robust

The ESP32 API reference

1 Like

Thanks mate! Hopefully I’ll get started with this project after my exams :slight_smile:

1 Like

audio amplifier

1 Like

Is there an actual benefit of using an ESP32 over an Arduino Nano? I only used an Arduino one time before, the ESP32 never. Plus the Arduino is cheaper on Amazon where I live :stuck_out_tongue: