piwho 1.3.2

Description:

piwho 1.3.2

Piwho is python wrapper around MARF
speaker recognition framework for the Raspberry pi and other SBCs. With
the Piwho you can implement speaker recognition in your projects.




Blink example
import RPi.GPIO as GPIO
import time
from piwho import recognition, vad

def blink(pin):
GPIO.setmode(GPIO.BOARD)
GPIO.setup(pin, GPIO.OUT)

for i in range(0,10):
GPIO.output(pin,GPIO.HIGH)
time.sleep(1)
GPIO.output(pin,GPIO.LOW)
time.sleep(1)

if __name__ == "__main__":
recog = recognition.SpeakerRecognizer('./')
vad.record()
name = recog.identify_speaker()
if name[0] == 'Abhishek':
blink(11)

Tested on

Ubuntu 15.10
Pi 1 model B (raspbian wheezy, jessie)
Pi 2 model B (raspbian wheezy, jessie, stretch)
CHIP



Installation
Update the Pi
$ sudo apt-get update
$ sudo apt-get upgrade
You need to have JDK (min version: 1.7) installed on your Pi.
# verify jdk is installed
$ java -version
Pyaudio is required to run audio recording script. (Optional)
# Install portaudio
$ sudo apt-get install portaudio19-dev
# Install python dev package
$ sudo apt-get install python2.7-dev
# Install pyaudio
$ pip install pyaudio
Piwho is on PyPI
$ pip install piwho
or clone the project from github
$ git clone https://www.github.com/Adirockzz95/Piwho.git
$ cd Piwho
$ python setup.py install


Tests
Tests are implemented using unittest framework:
$ pip install -r requirements.txt
$ python -m unittest discover -v ./tests


Documentation

Training the model
Recognition
Gender detection
integrating with Jasper



Tips / Caveats

Recognition/Training time depends on the length of an audio file.
If possible overclock your Pi- use Turbo mode.
Give maxmium RAM to CPU.
Read MARF
manual
to know how the it works.
Recognition speed is directly proportional to the CPU power.



Misc
Here are resources/similar projects I came across while working on this
project.

recognito : Works
very well on Pi, but data storage is not implemented.
voiceid
ALIZE
Shout
MARF
manual
LIUM
site



LICENSE
MIT