Infineon Sound of Propulsion by sop_hacker_bois

Inspiration

We wanted to learn something new and try out Infineon's track.

What it does

The app uses machine learning technology and the i2s microphone from Infineon mounted to a raspberry pi to distinguish between different vehicle types. How to use: You can record audio files with the Raspberry Pi via the web interface, which are then analyzed to determine which vehicle it is. You can also you the UI the gather more audio samples.

The tryout link provides a link to the model, not the actual app.

How we built it

Using sample data of different vehicle models from an online database we trained a machine learning model with Edge Impulse.

Challenges we ran into

Gathering quality audio data was more difficult than we thought. Furthermore it was quite challenging to distinguish different types of sound in general. As we came to a point where the expenses for the input overcome the generated output, we decided to switch from differentiating between petrol and diesel to differentiating between different vehicle types.

Accomplishments that we're proud of

We managed to get the raspberry pi up and running and also send sound data to the Edge Impulse. Also we improved our classification a lot. In the beginning all vehicle types looked almost the same, but after setting some more variables, we could differentiate a lot better.

What we learned

We learned more about machine learning and audio processing with the raspberry pi.