PyAudioVisualiser

Using scipy and pygame to create an audio visualiser which uses multiple consecutive fast fourier transforms to visualise an audio file.
Each fast fourier transform decomposes an interval of the audio signal into its constituent frequencies.

Getting Started

Variations
AudioVisualiserGUI.pyw

• Allows for GUI file name input
• Does not support copy and paste

AudioVisualiserGUI.exe and AudioVisualiserGUI_Entertainment.exe

• Available in the release
• Can be used without the required dependencies (apart from lame.exe for mp3 input)
• Doesn't allow customisation of variables :(
• Entertainment version centers the bars to make it look nicer

AudioVisualiserConsole.py

• Uses a command line for name input
• Works with linux and OS X (file path must be enclosed with quotation marks)

Writing Filepath
File path is written relative to the songs_file variable which defaults as '' (meaning the file path is relative to the main AudioVisualiser file).

If no extension of file path is given, the program assumes it to be wav.
If the file path ends with .mp3, the program attempts to convert it with lame.

NOTE: Python only natively accepts sound files encoded in .wav format
Download lame.exe and link its path to the AudioVisualiser for mp3 input.
By default, the program looks for lame in the same directory as itself.

Dependencies for .py version:

1. Python 3.7+
2. Scipy
3. Pygame
4. Pathlib

Python 3.7+ can be installed from

 https://www.python.org/downloads 

Scipy, Pygame and Pathlib are python libraries which can be installed using pip in the terminal (after downloading python 3.7+)

 pip install scipy pygame pathlib

Note: The GUI and EXE versions only work (properly) on windows

Customising the Visualiser

The first 8 variables can be edited to suit your viewing experience.

songs_file = ''              #Directory To Search For Songs :) [the path finding is relative to this]
lame_path = 'lame.exe'       #Path to lame.exe
screen_w                     #Screen Width
screen_h                     #Screen Height
percentage_displayed_f       #Percentage of frequencies to show (Removes higher frequencies) Range = [0, 1]
max_height_percentile        #Pecentile of amplitude that fills the entire height of screen Range = (0, 100]
fftlength                    #Number of samples per DFT (better to be a power of 2) 
                             #Longer fftlength results in greater frequency resolution but worse time resolution
entertainment                #Just for aesthetics
                             #Centers the lowest frequencies in the centers and higher ones at the end