Simple voice authentication using matplotlib, dtw, pyaudio and google firebase(work in progress). This uses matplotlib.pyplot to plot the two audio files spectrum the using dtw library which stands for Dynamic Time Warping. This then calculates the distance between the two plots of the audio files.

$ git clone https://github.com/greallye/voice_authentication_python.git
$ cd voice_authentication_python

Now we need to creat the folder to store audio files for users

$ mkdir audioRefrence

createv virtualenv to install all the libaries if you do not have virtualenv (pip install virtualenv)

$ virtualenv voiceauth

Then activate it

$ source voiceauth/bin/activate

Lets install all the libaries(will create a setup later when this project is further developed)

$ pip install pyaudio
$ pip install wave
$ pip install librosa
$ pip install matplotlib 
$ pip install dtw

Open the enrol.py update the config with firebase credntials Open the okVoice.py update the config with firebase credntials

In enrol.py change "Test Name" to the name you want to enrol.

First run the enrol.py

$ enrol.py

Once the console says Recording say a phrase you want to test again. You have 5 seconds.

Once recorded

$ okVoice.py

This will say recording again Say the same phrase that you enrolled with. It should return the name you enrolled with if it is a match. Record multiple people and test it out.