Voicher

implement a voice changer with python
this project is built on OSX
just for fun & still in progress...

Background Knowledge:

https://www.zhihu.com/question/29261034 http://www.cs.bu.edu/courses/cs101b1/slides/CS101.Lect28.Python.Audio.ppt.pdf

Format Conversion:

Convert audio file to .wav, process it, and then convert it to your output format.

sox nuo.mp3 nuo.wav
ffmpeg ...
mpg123 ...

Libraries:

PySoundFile:

for reading and writing audio files;
http://pysoundfile.readthedocs.org/en/0.8.1/
other libs: http://nbviewer.jupyter.org/github/mgeier/python-audio/blob/master/audio-files/index.ipynb

Python-SoundDevice:

for recording and playing audio;
http://python-sounddevice.readthedocs.org/en/0.3.1/
other libs: http://guzalexander.com/2012/08/17/playing-a-sound-with-python.html

Additional Notes:

use python2.7:

use pip2.7 install *
if sublime can not run successfully, just use terminal instead;

use python3.5:

download miniconda(python3.5) from http://conda.pydata.org/miniconda.html
manually configure .zshrc to modify PATH, if you use zsh
sublime python configuration: http://blog.shank.in/post/26276497763/sublime-text-how-to-change-the-version-of-python
may change the file "python" in miniconda's bin to another name, then "python3" refers to python3.5 while "python" refers to your previous version; sometimes, when using pip to install things, may need to rename it back;
matplotlib issue: http://stackoverflow.com/questions/21784641/installation-issue-with-matplotlib-python

Steps:

tests:

overlay two array;
with music and without music, cmp the diff or two graph;
compress elem then play; repeated elem then play;

non-real-time voice changer: "voice alteration algorithm" http://dsp-book.narod.ru/Pitch_shifting.pdf
real-time voice changer
learn more about the APIs