with open(filename, "wb") as f: f.write(total_data) # print(total_data.decode()) # total_data = total_data.decode() # total_data = total_data.split('+@+') # if len(total_data) == 2: # result = "N" # else: # result = "N" # print(result) # conn.send(result.encode()) #然后再发送数据 result1 = '' success = True flag = identifier[1] voiceprint_a = extraction.ertract_voiceprint(filename, sr=16000) if (flag == '0'): # save(identifier[0],voiceprint=voiceprint_a) voice = voiceprint_a print('注册') result1 = 'Y' else: print('identifier', identifier[0]) # voiceprint_e = findvoice(identifier[0]) Euclideandist = PairwiseDistance(2) print('voice', type(voice)) distance = Euclideandist(voice, voiceprint_a) print(distance) if distance <= 0.4: result1 = "Y"
filename_e = '/AttackSample/zzl1.wav' filename_a = '/AttackSample/liqi&&0.wav' y1,sr = librosa.load(filename_e,sr=16000,mono=True) y2,sr = librosa.load(filename_a,sr=16000,mono=True) print('y1:',len(y1)) print('y2:',len(y2)) #print('================') #y变成80000 #print(len(y1)) #with torch.no_grad(): # voiceprint_e = extraction.ertract_voiceprint(filename_e,sr=16000) # voiceprint_a = extraction.ertract_voiceprint(filename_a,sr=16000) #print(voiceprint_e) Euclideandist = PairwiseDistance(2) distance = 0 for i in range(20): voiceprint_e = extraction.ertract_voiceprint(filename_e, sr=16000) voiceprint_a = extraction.ertract_voiceprint(filename_a, sr=16000) distance += Euclideandist(voiceprint_e, voiceprint_a).item() print("distance:",distance/20) #fake_noise = attacker(voiceprint_e).squeeze() #while len(y2)>len(fake_noise): # fake_noise = np.concatenate((y1,y1),axis = 0)#y1 = y1 + y1 # y1 = y1[0:80000] # while len(y2)<80000: # y2 = np.concatenate((y2,y2),axis = 0) # #print(len(y2)) # y2 = y2[0:80000]
import InnerProduct import extraction import torch secu = InnerProduct.InnerProduct() voiceprint_a = extraction.ertract_voiceprint("Hjjgg&&0.wav", sr=16000) voiceprint_b = extraction.ertract_voiceprint("Hjjgg&&1.wav", sr=16000) A = torch.norm(voiceprint_a).item() B = torch.norm(voiceprint_b).item() print(voiceprint_a) print('torch.norm(voiceprint_a)', torch.norm(voiceprint_a)) print('A', A) print('type(A)', type(A)) a = voiceprint_a.cpu().numpy().tolist()[0] print(len(a)) print(type(a)) print('a', a) b = voiceprint_b.cpu().numpy().tolist()[0] b = list(b) print(len(b)) print(type(b)) print('b', b) print(max(b)) print(min(b)) s, C = secu.Step1(a) DSum = secu.Step2(b, C) innerproduct = secu.Step3(DSum, s) print('innerproduct', innerproduct) cos = innerproduct / (A * B) print('cos', cos)