from scipy.fftpack import fft
from scipy.io import wavfile # get the api
from scipy import arange
from python_speech_features import mfcc
from python_speech_features import delta
from python_speech_features import logfbank
import numpy as np
import glob
import os
import csv
from pydub import AudioSegment, scipy_effects


#  Chunking function 
def chunks(l, n):
    for i in xrange(0, len(l), n):
        yield l[i:i+n]

# Resampling function
def resample(arr, newLength):
    chunkSize = len(arr)/newLength
    return [np.mean(chunk) for chunk in chunks(arr, chunkSize)]


files = os.listdir("piano/wavs")
csvfile = open("output_piano_note_present.csv", "w")
csvfile1 = open("labels_piano_note_present.csv", "w")
csvfile2 = open("output_piano_note_not_present.csv", "w")
csvfile3 = open("labels_piano_note_not_present.csv", "w")
writer = csv.writer(csvfile, delimiter=',')
writer1 = csv.writer(csvfile1, delimiter=',')
writer2 = csv.writer(csvfile2, delimiter=',')
writer3 = csv.writer(csvfile3, delimiter=',')
counter = 0
for filename in files:
    if(".wav" not in filename):
        continue
    if(counter % 100 == 0):
        print counter
    filename1 = "piano/wavs/" + filename
    try:
        song = AudioSegment.from_wav(filename1)#(filename)
    except OSError:
        continue
    song = song.set_channels(1)
    y = song.get_array_of_samples()
    y = np.array(y)
    
    fs = song.frame_rate
    Ts = 1.0/fs
    t = arange(0,1,Ts)
    n = len(y)
    k = arange(n)
    T = n/fs  # where fs is the sampling frequency
    frqLabel = k/T
    frq = frqLabel[range(n/2)]
    frq = frq[:6000]
    Y = fft(y)/n # calculate fourier transform (complex numbers list)
    Y = Y[range(n/2)]  # you only need half of the fft list (real signal symmetry)
    Y = Y[:6000]
    if("C3" in filename):
        writer1.writerow("1")
        writer.writerow(abs(Y[::2]))
    else:
        writer3.writerow("0")
        writer2.writerow(abs(Y[::2]))
    counter += 1
csvfile.close()