WAV to Plot from scratch in Python. I know this can be done using SciPy or various other libraries but I wanted to learn more about wav data so I did this from scratch. Added bonus to doing this myself is that I could specify how accurately I wanted the data to be in order to speed up time on large files. Currently, this only works on 16bit mono…

wavplot.py

import os
import sys
import numpy as np
import struct
import warnings

path="sample.wav"
#get the file size in bytes
fileSize=os.path.getsize(path)  

f=open(path,"r")

#Go to ByteRate at offset 28
f.seek(28)

#Read in 4 bytes and convert to signed int
b = struct.unpack('i', f.read(4))
byteRate= b[0]

#Calculate duration of sound
ms=((fileSize-44)*1000)/byteRate

#Go to first byte of sound data
f.seek(44,0)

#secSize or Section Size will allow you to grab a "chunk" of 
#data and find the average instead of plotting every 
#individual point.Very useful for a quick analysis of 
#large files. 1 = 1:1 / 16 = 16:1
secSize = 16

dValue=0 #dataValue
dVals = [] #dataValues
mVals =[] #meanValues

done = 'n'
#Generate Y values
while (done != 'y') : 
	q=0
	while q < secSize :
		try:
			#Read 'short' from wav file
			dValue = struct.unpack('=h', f.read(2)) 
		except struct.error:
			#If your section size extends past the last byte
			#just break out of the loop
			done = 'y'
			break
		dVals.append(dValue[0])
		q += 1

	#Supress NumPys "Nan" warning
	warnings.simplefilter("ignore")

	#Find average of section and add it to mVals
	mVals.append(np.mean(dVals))
	dVals=[]
	dValue = 0

#Find the amount of time each 
#point will represent
xInc = round(float(ms)/len(mVals), 6)

x = 0
yInc = 1
y=0
pair=[]
pairs = []

#Generate X values for Y counterparts
while (x < ms):
	try:
		y = round(mVals[yInc], 2) 
	except IndexError:
		break
	
	pair.append(round(x, 6))
	pair.append(y)
	pairs.append(pair)
	x += xInc
	yInc += 1
	pair=[]

#Flush buffer just to be safe
sys.stdout.flush() 

#Print 'x,y' to console. 
#To write to file: ./ThisScript.py > data.txt
h = 0
while (h < len(pairs)):
	sys.stdout.write(str(pairs[h][0]) + ',' + str(pairs[h][1]) + '\n')
	#Uncomment next line for '(x,y)'
	#sys.stdout.write('(' + str(pairs[h][0]) + ',' + str(pairs[h][1]) + ')' + '\n')
	h+=1
f.close()