Pset6-Voice Recognition - Duplicate

import numpy as np import numpy.random as random from numpy.fft import fft from scipy.io import wavfile import matplotlib.pyplot as plt import seaborn as sns import os %matplotlib inline sns.set() sns.set(font_scale=1.5)

data_dir = './recordings/' # determine digits of interest (0 to 9) digits = [1,2] # change here to load more digits # dictionary that will store our values signals = {d:[] for d in digits} file_names = {d:[] for d in digits} # import files for filename in os.listdir(data_dir): # iterate over digits for d in digits: if filename.startswith(str(d)+'_'): wav = wavfile.read(data_dir+filename)[1] if len(wav.shape)<2: signals[d].append(wav) file_names[d].append(filename) # find maximum of vector length N = max([len(v) for d in digits for v in signals[d]])

# next we split our dataset in train and test # we will use a 80/20 random split. # create train/test split ix = np.arange(100) random.shuffle(ix) # select train entries ix_train = ix[:80] #select test entries ix_test = ix[80:]

# next we compute the average spectrum of each spoken digit in the training set. # we will consider a window up to 1.5 KHz # sampling rate is 8kHz Ts = 1.0/8000 ix_cut = int(np.ceil(1500*Ts*N)) # initialize dictionary for storing transforms transforms = {} # initialize dictionary for storing mean transforms mean_transforms = {} # compute mean transform of each digit and in the training set. # Make sure to only keep the spectrum up to 1.5kHz

# Code Solution to Q1 Here # Dictionary for non-normalized transforms transforms_not_normalized = {} for i in digits: # FFT dictionary for training set ix_train_fft = {} for j in ix_train: ix_train_fft[j] = fft(signals[i][j], n=N) ix_train_fft[j] = ix_train_fft[j][:ix_cut+1] # Find average spectral magnitude and place into correct dictionaries av_mag = [0] * (ix_cut+1) for j in ix_train: for k in range(ix_cut+1): av_mag[k] += abs(ix_train_fft[j][k])/80 transforms_not_normalized[i] = av_mag mean_transforms[i] = transforms_not_normalized[i]/ np.linalg.norm(transforms_not_normalized[i])

# In this next part, plot the average spectral magnitude of each digit.

# Code Solution to Q2 here %matplotlib inline xf = np.linspace(0, 1500, ix_cut+1) for d in [1,2]: y_axis = transforms_not_normalized[d] plt.plot(xf, y_axis) plt.show()

# classifier function # receives a vector, computes the product with average digits, and returns the max inner product # Input: sample x (vector) def mean_classifier(x): fft_x = fft(x, n=N)[:ix_cut+1] inn_prod = 0 similarity = {} for i in digits: for k in range(ix_cut): inn_prod += abs(fft_x[k]) * abs(mean_transforms[i][k]) similarity[i] = inn_prod inn_prod = 0 return max(similarity, key=similarity.get)

# Write anser for Q3b here

counter = [0] * len(digits) for i in digits: for j in ix_test: if i == mean_classifier(signals[i][j]): counter[i-1] += 1 for i in range(len(counter)): perc = counter[i]/20 * 100 digit_num = i+1 print("digit ", digit_num, " accuracy: ", perc, "%") # for i in counter: # print("digit ", counter[i+1], " accuracy: " counter_avg[i+1] "%")

# Write answer for Q4 here

%matplotlib inline data_dir = './recordings/' # determine digits of interest (0 to 9) digits = [0,1,2,3,4,5,6,7,8,9] # change here to load more digits # dictionary that will store our values signals = {d:[] for d in digits} file_names = {d:[] for d in digits} # import files for filename in os.listdir(data_dir): # iterate over digits for d in digits: if filename.startswith(str(d)+'_'): wav = wavfile.read(data_dir+filename)[1] if len(wav.shape)<2: signals[d].append(wav) file_names[d].append(filename) # find maximum of vector length N = max([len(v) for d in digits for v in signals[d]]) # next we split our dataset in train and test # we will use a 80/20 random split. # create train/test split ix = np.arange(100) random.shuffle(ix) # select train entries ix_train = ix[:80] #select test entries ix_test = ix[80:] # next we compute the average spectrum of each spoken digit in the training set. # we will consider a window up to 1.5 KHz # sampling rate is 8kHz Ts = 1.0/8000 ix_cut = int(np.ceil(1500*Ts*N)) # initialize dictionary for storing transforms transforms = {} # initialize dictionary for storing mean transforms mean_transforms = {} # compute mean transform of each digit and in the training set. # Make sure to only keep the spectrum up to 1.5kHz # Code Solution to Q1 Here # Dictionary for non-normalized transforms transforms_not_normalized = {} for i in digits: # FFT dictionary for training set ix_train_fft = {} for j in ix_train: ix_train_fft[j] = fft(signals[i][j], n=N) ix_train_fft[j] = ix_train_fft[j][:ix_cut+1] # Find average spectral magnitude and place into correct dictionaries av_mag = [0] * (ix_cut+1) for j in ix_train: for k in range(ix_cut+1): av_mag[k] += abs(ix_train_fft[j][k])/80 transforms_not_normalized[i] = av_mag mean_transforms[i] = transforms_not_normalized[i]/ np.linalg.norm(transforms_not_normalized[i]) xf = np.linspace(0, 1500, ix_cut+1) for d in digits: y_axis = transforms_not_normalized[d] plt.plot(xf, y_axis) plt.show() def mean_classifier(x): fft_x = fft(x, n=N)[:ix_cut+1] inn_prod = 0 similarity = {} for i in digits: for k in range(ix_cut): inn_prod += abs(fft_x[k]) * abs(mean_transforms[i][k]) similarity[i] = inn_prod inn_prod = 0 return max(similarity, key=similarity.get) counter = [0] * len(digits) for i in digits: for j in ix_test: if i == mean_classifier(signals[i][j]): counter[i-1] += 1 for i in range(len(counter)): perc = counter[i]/20 * 100 digit_num = i+1 print("digit ", digit_num, " accuracy: ", perc, "%") # for i in counter: # print("digit ", counter[i+1], " accuracy: " counter_avg[i+1] "%")

pip install -U notebook-as-pdf