• Clone entire directory into any folder of your choice
• Manually go through each file and change the absolute paths based on your file structure
• Inside the input_saver directory, run ./input.sh
• After the process finishes, check results/test.txt for the transcription of the recording
• the voskcall.py file is used for working with the vosk-transcriber and must be carefully altered
• This package is supposed to work with https://github.com/realaryann/Keyword-Select-Service (service to extract keywords)

Working pipeline of recording, transcribing, and storing

• sounddevice
• numpy
• scipy
• vosk

• Navigate to input_saver/find_device.py
• In a terminal, check the name of your sounddevice by running python3 -m sounddevice
• Replace the DEVICE_NAME variable with the name of the device

• Input_Saver currently takes string input from the command line
• Input Mapping
1. R: Record for 5 seconds
2. T: Record for 10 seconds
3. Y: Record for 20 seconds
4. Z: Exit program
• The program will continue looping until valid input is entered

• Navigate to input_saver/voskhear.py
• Replace the newf and filesrc variables (WARNING, could break everything!)

• Navigate to the parent directory (./input_saver
• Run ./clean.sh in ~/input_saver to clear out previous recordings and results

1. python3 inputaudio.py
2. python3 voskhear.py