hello2.py

from scikits.audiolab import wavread
from scikits.talkbox.features import mfcc
from scikits.talkbox.linpred.levinson_lpc import lpc
import os
import numpy as np
from test.LPC import LPCExtractor
from test.MFCC import get_mfcc_extractor
from collections import defaultdict
from sklearn.mixture import GMM
from collections import Counter
import pickle
import operator
import random
from random import randint

gmm_labels = []
gmm_instances = []
gmm_tables = {}
random_str = str(1290152)
f = open("pickle_dump/"+ random_str + "/ubm-pickle.txt","r")
ubm_model = pickle.load(f)
f.close()
f = open("pickle_dump/"+ random_str + "/correct_scores.txt","r")
correct_scores = pickle.load(f)
f.close()
base_dir = "/Users/adhithyar/Desktop/ece6255/"
l = LPCExtractor(16000)
mfcc_extractor = get_mfcc_extractor()
counter = 0
for file_name in os.listdir(base_dir + "pickle_dump/" + random_str + "/"):
    print file_name
    if file_name != random_str:
        counter = counter + 1
        f = open("pickle_dump/" + random_str + "/" + file_name)
        gmm_instance = pickle.load(f)
        f.close()
        k = file_name.split("-")[0]
        if file_name != "ubm-pickle.txt" or file_name != "correct_scores.txt":
            gmm_labels.append(k)
            gmm_tables[k] = gmm_instance
            gmm_instances.append(gmm_instance)

results = {}
f = open("testing_list"+random_str+".txt","r")
testing_set_filenames = pickle.load(f)
f.close()

base_dir = "/Users/adhithyar/Desktop/ece6255/converted_data/"
audio_files = os.listdir(base_dir)
avg = []
counter = 0
for file_name in audio_files:
    if file_name != "ubm":
        print file_name
        print counter + 1
        counter = counter + 1
        flits = file_name.split("-")
        data, fs = wavread(base_dir + file_name)[:2]
        ceps = mfcc_extractor.extract(data)
        lpcs = l.extract(data)
        b = np.concatenate((ceps, lpcs), axis=1)
        corresponding_gmm_instance = gmm_tables[flits[0]]
        orig_value = np.sum(corresponding_gmm_instance.score(b))
        print orig_value
        random_flit_entry = random.choice(gmm_labels)
        while random_flit_entry == flits[0]:
            random_flit_entry = random.choice(gmm_labels)
        while random_flit_entry == 'ubm':
            random_flit_entry = random.choice(gmm_labels)
        while random_flit_entry == "correct_scores.txt":
            random_flit_entry = random.choice(gmm_labels)
        corresponding_gmm_instance = gmm_tables[random_flit_entry]
        test_value = np.sum(corresponding_gmm_instance.score(b))
        print test_value
        avg.append(orig_value/test_value)

print avg
print min(avg)
print max(avg)
above_70_counter = 0
for value in avg:
    if value > 0.7:
        above_70_counter = above_70_counter + 1

print above_70_counter