This is a project for a model trained to detect a specific beep sound in pre-loaded audio files or through microphone input from the external environment.
- pip install sounddevice
- pip install librosa
- pip install matplotlib
- pip install numpy
- pip install pandas
- pip install tensorflow
- pip install sklearn
- pip install pyttsx3
- pip install playsound==1.2.2
Install any other required dependecies on raspberry pi.
Follow instructions on this link to install FFmpeg to the root directory of your computer if you are on windows OS. https://www.wikihow.com/Install-FFmpeg-on-Windows FFmpeg helps librosa to decode other formats.
If Linux OS: run this command "apt install ffmpeg"
- Collect sample data and background data using PreparingData.py
- Preprocess the collected data using Preprocessing.py
- Train the model on the collected data
- Run test.py to test trained model with pre-loaded wav files in the Audio_database_dataset csv file. The program will first slice the selected WAV file into smaller wav files and iterate through these slices trying to detect the beep. For every beep detected, the program will play the beep sound and turn the servo motor 10 degrees and then back to 0 degrees.
- Run main.py when you want to test the model with inputs from a microphone connected to your PC or to your Raspberry pi. The program will continuously listen to input from the mic, to try and detect the beep sound.
Happy Coding!!!