#!/usr/bin/python

#

# stjepan.henc@fer.hr

# 25.11.2013

#

import wave, struct, math, sys, time

import matplotlib.pylab as pl

import scipy

import scipy.fftpack

import pylab

import scipy.stats

import numpy

#audio = wave.open("ispitivanje.wav", mode="r")

audio = wave.open("smallest.wav", mode="r")

length = audio.getnframes()

sample_rate = audio.getframerate()

desni = [0] * length

proba = [0] * length

# read & preprocess the data from the file

for i in range(0, length):

data = audio.readframes(1)

tmp = struct.unpack("<hh", data)

#ljevi[i] = tmp[0]

desni[i] = tmp[1]

stanje = False # je li zvucno ili tiho

brojaci = []

tisina_int = []

zvuk_int = []

last = 0

#for k in range (5, 20):

treshold = 20

frame = 50 # dobiveni brute-force optimizacijom

brojac = 0

for i in range(frame, length):

bul = True

for j in range(0, frame - 1):

if abs(desni[i - j]) > treshold:

bul = False

if bul == True and stanje == False: # onda treba promijeniti stanje

stanje = True

proba[i - frame] = 32000

brojac = brojac + 1

tisina_int.append(1.0 * (i - frame - last) / sample_rate)

last = i - frame

if bul == False and stanje == True: # onda treba promijeniti stanje

stanje = False

proba[i] = -32000

brojac = brojac + 1

zvuk_int.append(1.0 * (i - last) / sample_rate)

last = i

brojaci.append(brojac)

audio.close()

print(brojac)

#arr = numpy.array([tisina_int, zvuk_int])

tisina_sred = numpy.mean(tisina_int)

print(tisina_sred)

tisina_dev = numpy.std(tisina_int)

print(tisina_dev)

zvuk_sred = numpy.mean(zvuk_int)

print(zvuk_sred)

zvuk_dev = numpy.std(zvuk_int)

print(zvuk_dev)

brojevi = pl.linspace(5, 19, 15)

time = pl.linspace(1, length, length)

pl.subplot(211)

pl.plot(time, desni)

pl.subplot(212)

pl.plot(time, proba)

pl.show()

#TODO

# odvajanje intervala

# aproksimacija sinusoide