voiceClassifier

A simple package for voice classifiaction


License
MIT
Install
pip install voiceClassifier==1.1.2.1

Documentation

voiceClassifier

A module for classifying voice samples. The layout/Class diagram of package is as follows.

Package Layout

Installation:

Run this command in terminal

pip install voiceclassifier

This package has dependancy on librosa and other audio related packages. Please check the requirement.txt for further details.

Please note that this package is extensively tested on Linux enviourment(Ubuntu.)

Examples:

Let's have a look at some examples to gain more understanding.

1. Trainning on your own data :

  • For this you can simply run evaluate.py file from the package if you downloaded zip from github. For this please keep your voice_samples ready. The name of the voice file will be considerd as a class label.Pass list of all your voice_samples path when asked. Train Samples

  • If you have downloaded the package using pip, then you can train as follow...

from voiceClassifier import evaluate
evaluate.train(
               ["/home/rshinde/Desktop/data/Shekhar.m4a","/home/rshinde/Desktop/data/Shubham.m4a",
                "/home/rshinde/Desktop/data/Dhruvesh.m4a","/home/rshinde/Desktop/data/Rohit.m4a"]

2. classifying the voice sample

Herewith the code...

from voiceClassifier.voiceClassify import VoiceClassify

vc = VoiceClassify()
# to get classes
print(vc.classes_())

# pass voice_samples for actual prediciton...
prediciton,probabilities,classes_ = vc.predict("/home/rshinde/Desktop/Dhruvesh.wav")

Output

Audio preprocessed..
(array(['Dhruvesh'], dtype=object),
 array([[0.55461039, 0.05390575, 0.03683959, 0.35464427]]),
 array(['Dhruvesh', 'Rohit', 'Shekhar', 'Shubham'], dtype=object))